plant tissue culture engineering

Transcription

PLANT TISSUE CULTURE ENGINEERING
FOCUS ON BIOTECHNOLOGY
Volume 6
Series Editors
MARCEL HOFMAN
Centre for Veterinary and Agrochemical Research, Tervuren, Belgium
JOZEF ANNÉ
Rega Institute, University of Leuven, Belgium
Volume Editors
S. DUTTA GUPTA
Department of Agricultural and Food Engineering,
Indian Institute of Technology,
Kharagpur, India
YASUOMI IBARAKI
Department of Biological Science,
Yamaguchi University,
Yamaguchi, Japan
COLOPHON
Focus on Biotechnology is an open-ended series of reference volumes produced by
Springer in co-operation with the Branche Belge de la Société de Chimie Industrielle
a.s.b.l.
The initiative has been taken in conjunction with the Ninth European Congress on
Biotechnology. ECB9 has been supported by the Commission of the European
Communities, the General Directorate for Technology, Research and Energy of the
Wallonia Region, Belgium and J. Chabert, Minister for Economy of the Brussels Capital
Region.
Plant Tissue Culture Engineering
Edited by
S. DUTTA GUPTA
Department of Agricultural and Food Engineering,
Indian Institute of Technology,
Kharagpur, India
and
YASUOMI IBARAKI
Department of Biological Science,
Yamaguchi University,
Yamaguchi, Japan
A C.I.P. Catalogue record for this book is available from the Library of Congress.
ISBN 978-1-4020-3594-4 (HB)
ISBN 978-1-4020-3694-1 (e-book)
Published by Springer,
P.O. Box 17, 3300 AA Dordrecht, The Netherlands.
www.springer.com
Printed on acid-free paper
First edition 2006
Reprinted 2008
All Rights Reserved
© 2008 Springer
No part of this work may be reproduced, stored in a retrieval system, or transmitted
in any form or by any means, electronic, mechanical, photocopying, microfilming, recording
or otherwise, without written permission from the Publisher, with the exception
of any material supplied specifically for the purpose of being entered
and executed on a computer system, for exclusive use by the purchaser of the work.
FOREWORD
It is my privilege to contribute the foreword for this unique volume entitled: “Plant
Tissue Culture Engineering,” edited by S. Dutta Gupta and Y. Ibaraki. While there have
been a number of volumes published regarding the basic methods and applications of
plant tissue and cell culture technologies, and even considerable attention provided to
bioreactor design, relatively little attention has been afforded to the engineering
principles that have emerged as critical contributions to the commercial applications of
plant biotechnologies. This volume, “Plant Tissue Culture Engineering,” signals a
turning point: the recognition that this specialized field of plant science must be
integrated with engineering principles in order to develop efficient, cost effective, and
large scale applications of these technologies.
I am most impressed with the organization of this volume, and the extensive list of
chapters contributed by expert authors from around the world who are leading the
emergence of this interdisciplinary enterprise. The editors are to be commended for
their skilful crafting of this important volume. The first two parts provide the basic
information that is relevant to the field as a whole, the following two parts elaborate on
these principles, and the last part elaborates on specific technologies or applications.
Part 1 deals with machine vision, which comprises the fundamental engineering
tools needed for automation and feedback controls. This section includes four chapters
focusing on different applications of computerized image analysis used to monitor
photosynthetic capacity of micropropagated plants, reporter gene expression, quality of
micropropagated or regenerated plants and their sorting into classes, and quality of cell
culture proliferation. Some readers might be surprised by the use of this topic area to
lead off the volume, because many plant scientists may think of the image analysis tools
as merely incidental components for the operation of the bioreactors. The editors
properly focus this introductory section on the software that makes the real differences
in hardware performance and which permits automation and efficiency.
As expected the larger section of the volume, Part 2 covers Bioreactor Technologythe hardware that supports the technology. This section includes eight chapters
addressing various applications of bioreactors for micropropagation, bioproduction of
proteins, and hairy root culture for production of medicinal compounds. Various
engineering designs are discussed, along with their benefits for different applications,
including airlift, thin-film, nutrient mist, temporary immersion, and wave bioreactors.
These chapters include discussion of key bioprocess control points and how they are
handled in various bioreactor designs, including issues of aeration, oxygen transport,
nutrient transfer, shear stress, mass/energy balances, medium flow, light, etc.
Part 3 covers more specific issues related to Mechanized Micropropagation. The two
chapters in this section address the economic considerations of automated
micropropagation systems as related to different types of tissue proliferation, and the
use of robotics to facilitate separation of propagules and reduce labour costs. Part 4,
Engineering Cultural Environment, has six chapters elaborating on engineering issues
related to closed systems, aeration, culture medium gel hardness, dissolved oxygen,
v
Foreword
photoautotrophic micropropagation and temperature distribution inside the culture
vessel.
The last part (Part 5) includes four chapters that discuss specific applications in
Electrophysiology, Ultrasonics, and Cryogenics. Benefits have been found in the use of
both electrostimulation and ultrasonics for manipulation of plant regeneration.
Electrostimulation may be a useful tool for directing signal transduction within and
between cells in culture. Ultrasound has also applications in monitoring tissue quality,
such as state of hyperhydricity. Finally the application of engineering principles has
improved techniques and hardware used for long-term cryopreservation of plant stock
materials.
Readers of this volume will find a unique collection of chapters that will focus our
attention on the interface of plant biotechnologies and engineering technologies. I look
forward to the stimulation this volume will bring to our colleagues and to this emerging
field of research and development!
Gregory C. Phillips, Ph. D.
Dean, College of Agriculture
Arkansas State University
vi
PREFACE
Plant tissue culture has now emerged as one of the major components of plant
biotechnology. This field of experimental botany begins its journey with the concept of
‘cellular totipotency’ for demonstration of plant morphogenesis. Decades of research in
plant tissue culture has passed through many challenges, created new dreams and
resulted in landmark achievements. Considerable progress has been made with regard to
the improvement of media formulations and techniques of cell, tissue, organ, and
protoplast culture. Such advancement in cultural methodology led many recalcitrant
plants amenable to in vitro regeneration and to the development of haploids, somatic
hybrids and pathogen free plants. Tissue culture methods have also been employed to
study the basic aspects of plant growth, metabolism, differentiation and morphogenesis
and provide ideal opportunity to manipulate these processes.
Recent development of in vitro techniques has demonstrated its application in rapid
clonal propagation, regeneration and multiplication of genetically manipulated superior
clones, production of secondary metabolites and ex-situ conservation of valuable
germplasms. This has been possible not only due to the refinements of cultural practices
and applications of cutting-edge areas of molecular biology but also due to the judicious
inclusion of engineering principles and methods to the system. In the present scenario,
inclusion of engineering principles and methods has transformed the fundamental in
vitro techniques into commercially viable technologies. Apart from the
commercialization of plant tissue culture, engineering aspects have also made it
possible to improve the regeneration of plants and techniques of cryopreservation.
Strategies evolved utilize the disciplines of chemical, mechanical, electrical, cryogenics,
and computer science and engineering.
In the years to come, the application of plant tissue culture for various
biotechnological purposes will increasingly depend on the adoption of engineering
principles and better understanding of their interacting factors with biological system.
The present volume provides a cohesive presentation of the engineering principles and
methods which have formed the keystones in practical applications of plant tissue
culture, describes how application of engineering methods have led to major advances
in commercial tissue culture as well as in understanding fundamentals of
morphogenesis and cryopreservation, and focuses directions of future research, as we
envisage them. We hope the volume will bridge the gap between conventional plant
tissue culturists and engineers of various disciplines.
A diverse team of researchers, technologists and engineers describe in lucid manner
how various engineering disciplines contribute to the improvement of plant tissue
culture techniques and transform it to a technology. The volume includes twenty four
chapters presenting the current status, state of the art, strength and weaknesses of the
strategy applicable to the in vitro system covering the aspects of machine vision,
bioreactor technology, mechanized micropropagation, engineering cultural environment
and physical aspects of plant tissue engineering. The contributory chapters are written
by international experts who are pioneers, and have made significant contributions to
vii
Preface
this emerging interdisciplinary enterprise. We are indebted to the chapter contributors
for their kind support and co-operation. Our deepest appreciation goes to Professor G.C.
Phillips for sparing his valuable time for writing the Foreword. We are grateful to
Professor Marcel Hofman, the series editor, ‘Focus on Biotechnology’ for his critical
review and suggestions during the preparation of this volume.
Our thanks are also due to Dr. Rina Dutta Gupta for her efforts in checking the
drafts and suggesting invaluable clarifications. We are also thankful to Mr. V.S.S.
Prasad for his help during the preparation of camera ready version. Finally, many thanks
to Springer for their keen interest in bringing out this volume in time with quality work.
S. Dutta Gupta
Y. Ibaraki
Kharagpur/Yamaguchi, January 2005
viii
TABLE OF CONTENTS
FOREWORD………………………………………………………………………. ….v
PREFACE…………………………………………………………………………. …vii
TABLE OF CONTENTS………………………………………………………………1
PART 1...................................................................................................................... 13
MACHINE VISION.................................................................................................. 13
Evaluation of photosynthetic capacity in micropropagated plants by image
analysis ................................................................................................................. 15
Yasuomi Ibaraki .................................................................................................... 15
1. Introduction ................................................................................................... 15
2. Basics of chlorophyll fluorescence ............................................................... 16
3. Imaging of chlorophyll fluorescence for micropropagated plants................ 18
3.1. Chlorophyll fluorescence in in vitro cultured plants.............................. 18
3.2. Imaging of chlorophyll fluorescence ..................................................... 21
3.3. Imaging of chlorophyll fluorescence in micropropagated plants .......... 22
4. Techniques for image-analysis-based evaluation of photosynthetic capacity 25
5. Estimation of light distribution inside culture vessels .................................. 26
5.1. Understanding light distribution in culture vessels................................ 26
5.2. Estimation of light distribution within culture vessels .......................... 26
6. Concluding remarks ...................................................................................... 27
References ......................................................................................................... 28
Monitoring gene expression in plant tissues ..................................................... 31
John J. Finer, Summer L. Beck, Marco T. Buenrostro-Nava, Yu-Tseh Chi and
Peter P. Ling .......................................................................................................... 31
1. Introduction ................................................................................................... 31
2. DNA delivery ................................................................................................ 32
2.1. Particle bombardment ............................................................................ 32
2.2. Agrobacterium........................................................................................ 33
3. Transient and stable transgene expression .................................................... 33
4. Green fluorescent protein .............................................................................. 34
4.1. GFP as a reporter gene ........................................................................... 34
4.2. GFP image analysis................................................................................ 35
4.3. Quantification of the green fluorescence protein in vivo ....................... 36
5. Development of a robotic GFP image acquisition system............................ 37
5.1. Overview ................................................................................................ 37
5.2. Robotics platform................................................................................... 37
5.3. Hood modifications ................................................................................ 39
5.4. Microscope and camera.......................................................................... 40
5.5. Light source and microscope optics....................................................... 40
6. Automated image analysis ............................................................................ 41
6.1. Image registration................................................................................... 41
6.2. Quantification of GFP ............................................................................ 43
1
Table of Contents
7. Conclusions ...................................................................................................
Acknowledgements ...........................................................................................
References .........................................................................................................
Applications and potentials of artificial neural networks in plant tissue
culture ..................................................................................................................
V.S.S. Prasad and S. Dutta Gupta .........................................................................
1. Introduction ...................................................................................................
2. Artificial neural networks..............................................................................
2.1. Structure of ANN ...................................................................................
2.2. Working principle and properties of ANN.............................................
2.2.1. Computational property of a node...................................................
2.2.2. Training mechanisms of ANN ........................................................
2.3. Types of artificial neural networks ........................................................
2.3.1. Classification and clustering models...............................................
2.3.2. Association models .........................................................................
2.3.3. Optimization models .......................................................................
2.3.4. Radial basis function networks (RBFN) .........................................
2.4. Basic strategy for network modelling ....................................................
2.4.1. Database ..........................................................................................
2.4.2. Selection of network structure ........................................................
2.4.2.1. Number of input nodes.............................................................
2.4.2.2. Number of hidden units............................................................
2.4.2.3. Learning algorithm...................................................................
2.4.3. Training and validation of the network...........................................
3. Applications of ANN in plant tissue culture systems ...................................
3.1. In vitro growth simulation of alfalfa ......................................................
3.2. Classification of plant somatic embryos ................................................
3.3. Estimation of biomass of plant cell cultures ..........................................
3.4. Simulation of temperature distribution inside a plant culture vessel.....
3.5. Estimation of length of in vitro shoots...................................................
3.6. Clustering of in vitro regenerated plantlets into groups.........................
4. Conclusions and future prospects..................................................................
Acknowledgement.............................................................................................
References .........................................................................................................
Evaluation of plant suspension cultures by texture analysis...........................
Yasuomi Ibaraki ....................................................................................................
1. Introduction ...................................................................................................
2. Microscopic and macroscopic image uses in plant cell suspension culture .
3. Texture analysis for macroscopic images of cell suspensions......................
3.1. Texture features......................................................................................
3.2. Texture analysis for biological objects ..................................................
3.3. Texture analysis for cell suspension culture ..........................................
3.4. Considerations for application of texture analysis.................................
4. Evaluation of embryogenic potential of cultures by texture analysis ...........
4.1. Evaluation of embryogenic potential of cultures ...................................
4.2. Texture analysis based evaluation of embryogenic potential ................
2
43
44
44
47
47
47
48
48
49
49
51
51
51
52
52
52
52
52
53
54
54
54
55
56
56
58
58
59
61
61
65
66
66
69
69
69
69
71
71
72
73
73
73
73
74
Table of Contents
5. Concluding remarks ...................................................................................... 77
References ......................................................................................................... 77
PART 2...................................................................................................................... 81
BIOREACTOR TECHNOLOGY ............................................................................. 81
Bioengineering aspects of bioreactor application in plant propagation ........ 83
Shinsaku Takayama and Motomu Akita ............................................................... 83
1. Introduction ................................................................................................... 83
2. Advantages of the use of bioreactor in plant propagation ............................ 84
3. Agar culture vs. liquid culture....................................................................... 85
4. Transition from shake culture to bioreactor culture...................................... 85
5. Types of bioreactors for plant propagation ................................................... 86
6. Preparation of propagules for inoculation to bioreactor ............................... 87
7. Characteristics of bioreactor for plant propagation....................................... 88
7.1. Fundamental configuration of bioreactor............................................... 88
7.2. Aeration and medium flow characteristics............................................. 90
7.2.1. Medium flow characteristics ........................................................... 90
7.2.2. Medium mixing ............................................................................... 91
7.2.3. Oxygen demand and oxygen supply ............................................... 92
7.3. Light illumination and transmittance ..................................................... 93
8. Examples of bioreactor application in plant propagation ............................. 95
9. Aseptic condition and control of microbial contamination........................... 95
10. Scale-up to large bioreactor......................................................................... 96
10.1. Propagation of Stevia shoots in 500 L bioreactor ................................ 96
10.2. Safe inoculation of plant organs into bioreactor .................................. 98
11. Prospects...................................................................................................... 98
References ......................................................................................................... 98
Agitated, thin-films of liquid media for efficient micropropagation............ 101
Jeffrey Adelberg .................................................................................................. 101
1. Introduction ................................................................................................. 101
2. Heterotrophic growth and nutrient use........................................................ 102
2.1. Solutes in semi-solid agar .................................................................... 102
2.2. Solutes in stationary liquids ................................................................. 103
2.3. Sugar in shaker flasks and bioreactors ................................................. 105
3. Efficiency in process ................................................................................... 108
3.1. Shoot morphology for cutting and transfer process ............................. 108
3.2. Space utilization on culture shelf ......................................................... 109
3.3. Plant quality.......................................................................................... 109
4. Vessel and facility design............................................................................ 110
4.1. Pre-existing or custom designed vessel ............................................... 110
4.2. Size and shape ...................................................................................... 111
4.3. Closures and ports ................................................................................ 112
4.4. Biotic contaminants.............................................................................. 113
4.5. Light and heat....................................................................................... 113
5. Concluding remarks .................................................................................... 115
Disclaimer ....................................................................................................... 115
References ....................................................................................................... 115
3
Table of Contents
Design, development, and applications of mist bioreactors for
micropropagation and hairy root culture .......................................................
Melissa J. Towler, Yoojeong Kim, Barbara E. Wyslouzil,
Melanie J. Correll, and Pamela J. Weathers .....................................................
1. Introduction .................................................................................................
2. Mist reactor configurations .........................................................................
3. Mist reactors for micropropagation.............................................................
4. Mist reactors for hairy root culture .............................................................
5. Mist deposition modelling...........................................................................
6. Conclusions .................................................................................................
Acknowledgements .........................................................................................
References .......................................................................................................
Bioreactor engineering for recombinant protein production using
plant cell suspension culture ...........................................................................
Wei Wen Su.........................................................................................................
1. Introduction .................................................................................................
2. Culture characteristics .................................................................................
2.1. Cell morphology, degree of aggregation, and culture rheology ..........
2.2. Foaming and wall growth.....................................................................
2.3. Shear sensitivity ...................................................................................
2.4. Growth rate, oxygen demand, and metabolic heat loads .....................
3. Characteristics of recombinant protein expression .....................................
4. Bioreactor design and operation..................................................................
4.1. Bioreactor operating strategies.............................................................
4.2. Bioreactor configurations and impeller design ....................................
4.3. Advances in process monitoring ..........................................................
5. Future directions..........................................................................................
References .......................................................................................................
Types and designs of bioreactors for hairy root culture ...............................
Yong-Eui Choi, Yoon-Soo Kim and Kee-Yoeup Paek.......................................
1. Introduction .................................................................................................
2. Advantage of hairy root cultures.................................................................
3. Induction of hairy roots ...............................................................................
4. Large-scale culture of hairy roots ...............................................................
4.1. Stirred tank reactor ...............................................................................
4.2. Airlift bioreactors .................................................................................
4.3. Bubble column reactor .........................................................................
4.4. Liquid-dispersed bioreactor .................................................................
5. Commercial production of Panax ginseng roots via balloon
type bioreactor ...............................................................................................
References .......................................................................................................
Oxygen transport in plant tissue culture systems ..........................................
Wayne R. Curtis and Amalie L. Tuerk................................................................
1. Introduction .....................................................................................................
4
119
119
119
120
122
125
128
130
131
131
135
135
135
136
137
140
141
145
146
148
148
151
153
154
155
155
161
161
161
162
162
163
164
164
165
165
166
169
169
173
173
173
Table of Contents
2. Intraphase transport .....................................................................................
2.1. Oxygen transport in the gas phase .......................................................
2.2. Oxygen transport in the liquid phase ...................................................
2.3. Oxygen transport in solid (tissue) phase ..............................................
3. Interphase transport .....................................................................................
3.1. Oxygen transport across the gas-liquid interface.................................
3.2. Oxygen transport across the gas-solid interface ..................................
3.3. Oxygen transport across the solid-liquid interface ..............................
4. Example: oxygen transport during seed germination in aseptic liquid
culture .............................................................................................................
4.1. The experimental system used for aseptic germination of seeds in
liquid culture................................................................................................
4.2. Experimental observation of oxygen limitation...................................
4.3. Characterization of oxygen mass transfer ............................................
5. Conclusions .................................................................................................
References .......................................................................................................
Temporary immersion bioreactor ...................................................................
F. Afreen..............................................................................................................
1. Introduction .................................................................................................
2. Requirement of aeration in bioreactor: mass oxygen transfer ....................
3. Temporary immersion bioreactor................................................................
3.1. Definition and historical overview.......................................................
3.2. Design of a temporary immersion bioreactor ......................................
3.3. Advantages of temporary immersion bioreactor..................................
3.4. Scaling up of the system: temporary root zone immersion bioreactor
3.5. Design of the temporary root zone immersion bioreactor ...................
3.6. Case study – photoautotrophic micropropagation of coffee ................
3.7. Advantages of the system.....................................................................
4. Conclusions .................................................................................................
References .......................................................................................................
Design and use of the wave bioreactor for plant cell culture ........................
Regine Eibl and Dieter Eibl ................................................................................
1. Introduction .................................................................................................
2. Background .................................................................................................
2.1. Disposable bioreactor types for in vitro plant cultures ........................
2.2. The wave: types and specification .......................................................
3. Design and engineering aspects of the wave...............................................
3.1. Bag design ............................................................................................
3.2. Hydrodynamic characterisation ...........................................................
3.3. Oxygen transport efficiency .................................................................
4. Cultivation of plant cell and tissue cultures in the wave.............................
4.1. General information .............................................................................
4.2. Cultivation of suspension cultures .......................................................
4.3. Cultivation of hairy roots .....................................................................
4.4. Cultivation of embryogenic cultures....................................................
5
175
175
176
177
179
179
179
180
181
181
182
182
185
185
185
187
187
187
188
189
189
189
190
191
191
193
198
199
200
203
203
203
204
204
206
209
209
210
217
217
217
220
222
223
Table of Contents
5. Conclusions .................................................................................................
References .......................................................................................................
PART 3....................................................................................................................
MECHANIZED MICROPROPAGATION ............................................................
Integrating automation technologies with commercial micropropagation .
Carolyn J. Sluis....................................................................................................
1. Introduction .................................................................................................
2. Biological parameters..................................................................................
2.1. The plant’s growth form affects mechanized handling........................
2.2. Microbial contaminants hinder scale-up ..............................................
3. Physical parameters.....................................................................................
3.1. Culture vessels......................................................................................
3.2. Physical orientation of explants for subculture or singulation.............
3.3. Gas phase of the culture vessel impacts automation............................
4. Economic parameters ..................................................................................
4.1. Baseline cost models ............................................................................
4.2. Economics of operator-assist strategies ...............................................
4.3. Organization of the approach to rooting: in vitro or ex vitro ...............
4.4. Economics of new technologies...........................................................
5. Business parameters ....................................................................................
5.1. Volumes per cultivar ............................................................................
5.2. Seasons .................................................................................................
5.3. Cost reduction targets...........................................................................
6. Political parameters .....................................................................................
7. Conclusions .................................................................................................
References .......................................................................................................
Machine vision and robotics for the separation and regeneration of plant
tissue cultures.....................................................................................................
Paul H. Heinemann and Paul N. Walker.............................................................
1. Introduction .................................................................................................
2. Examples of automation and robotics .........................................................
3. Robotic system component considerations .................................................
3.1. Plant growth systems for robotic separation ........................................
3.1.1. Nodes.............................................................................................
3.1.2. Clumps ..........................................................................................
3.2. An experimental shoot identification system for shoot clumps...........
3.2.1. Shoot identification using the Arc method ...................................
3.2.2. Shoot identification using the Hough transform method..............
3.2.3. Testing the Hough transform ........................................................
3.3. Robotic mechanisms for shoot separation ...........................................
3.3.1. Manual separation device..............................................................
3.3.2. Automated separation device ........................................................
3.3.3. Single image versus real-time imaging for shoot separation ........
3.3.4. Shoot re-growth.............................................................................
6
224
224
224
229
229
231
231
231
232
232
235
236
237
237
238
238
238
241
241
242
242
243
244
244
246
247
248
248
253
253
253
253
254
255
255
255
256
257
259
263
264
264
265
268
269
Table of Contents
3.3.5. Cycle time .....................................................................................
3.3.6. Commercial layout ........................................................................
References .......................................................................................................
PART 4....................................................................................................................
ENGINEERING CULTURAL ENVIRONMENT .................................................
Closed systems for high quality transplants using minimum resources ......
T. Kozai ...............................................................................................................
1. Introduction .................................................................................................
2. Why transplant production systems? ..........................................................
3. Why closed systems? ..................................................................................
4. Commercialization of closed transplant production systems......................
5. General features of high quality transplants ...............................................
6. Sun light vs. use of lamps as light source in transplant production............
7. Closed plant production system ..................................................................
7.1. Definition .............................................................................................
7.2. Main components .................................................................................
7.3. Characteristics of main components of the closed system...................
7.4. Equipments and facilities: a comparison .............................................
7.5. Features of the closed system vs. greenhouse......................................
7.6. Equality in Initial investment ...............................................................
7.7. Reduction in costs for transportation and labour .................................
7.8. Uniformity and precise control of microenvironment .........................
7.9. Growth, development and uniformity of transplants ...........................
8. Value-added transplant production in the closed system............................
8.1. Tomato (Lycopersicon esculentum Mill.) ............................................
8.2. Spinach (Spinacia oleracea) ................................................................
8.3. Sweet potato (Ipomoea batatas L. (Lam.)) ..........................................
8.4. Pansy (Viola x wittrockiana Gams.).....................................................
8.5. Grafted transplants ...............................................................................
8.6. Vegetable transplants for field cultivation ...........................................
9. Increased productivity to that of the greenhouse ........................................
10. Costs for heating, cooling, ventilation and CO2 enrichment.....................
10.1. Heating cost........................................................................................
10.2. Cooling load and electricity consumption .........................................
10.3. Cooling cost........................................................................................
10.4. Electricity consumption .....................................................................
10.5. Electricity cost is 1-5% of sales price of transplants .........................
10.6. Relative humidity ...............................................................................
10.7. Par utilization efficiency ....................................................................
10.8. Low ventilation cost ...........................................................................
10.9. CO2 cost is negligibly small ...............................................................
10.10. Water requirement for irrigation ......................................................
10.11. Disinfection of the closed system is easy.........................................
10.12. Simpler environmental control unit .................................................
10.13. Easier production management ........................................................
10.14. The closed system is environment friendly......................................
7
270
270
271
273
273
275
275
275
276
278
280
280
282
284
284
284
285
285
286
290
291
292
293
293
294
295
295
297
297
298
299
300
300
301
301
303
303
304
304
305
305
306
307
307
308
308
Table of Contents
10.15. The closed system is safer................................................................
11. Conclusion.................................................................................................
Acknowledgement...........................................................................................
References .......................................................................................................
Aeration in plant tissue culture........................................................................
S.M.A. Zobayed ..................................................................................................
1. Introduction .................................................................................................
2. Principles of aeration in tissue culture vessel .............................................
2.1. Aeration by bulk flow ..........................................................................
2.2. Aeration by diffusion ...........................................................................
2.3. Humidity-induced convection in a tissue culture vessel......................
2.4. Aeration by venturi-induced convection..............................................
2.5. Forced aeration by mass flow ..............................................................
3. Conclusions .................................................................................................
References .......................................................................................................
Tissue culture gel firmness: measurement and effects on growth................
Stewart I. Cameron..............................................................................................
1. Introduction .................................................................................................
2. Measurement of gel hardness......................................................................
3. Gel hardness and pH ...................................................................................
4. The dynamics of syneresis ..........................................................................
5. Conclusion...................................................................................................
References .......................................................................................................
Effects of dissolved oxygen concentration on somatic embryogenesis .........
Kenji Kurata and Teruaki Shimazu.....................................................................
1. Introduction .................................................................................................
2. Relationship between DO concentration and somatic embryogenesis .......
2.1. Culture system and DO concentration variations ................................
2.2. Time course of the number of somatic embryos..................................
2.3. Relationship between somatic embryogenesis and oxygen
concentration...............................................................................................
3. Dynamic control of DO concentration to regulate torpedo-stage embryos
3.1. The method of dynamic DO control ....................................................
3.2. Results of dynamic DO control............................................................
4. Conclusions .................................................................................................
References .......................................................................................................
A commercialized photoautotrophic micropropagation system...................
T. Kozai and Y. Xiao...........................................................................................
1. Introduction .................................................................................................
2. Photoautotrophic micropropagation............................................................
2.1. Summary of our previous work............................................................
3. The PAM (photoautotrophic micropropagation) system and its
components......................................................................................................
3.1. System configuration............................................................................
3.2. Multi-shelf unit.....................................................................................
3.3. Culture vessel unit................................................................................
8
309
310
311
311
313
313
313
314
317
319
321
325
326
326
327
329
329
329
330
333
334
335
336
339
339
339
341
341
342
346
347
347
351
352
352
355
355
355
356
356
357
357
358
360
Table of Contents
3.4. Forced ventilation unit for supplying CO2-enriched air....................... 360
3.5. Lighting unit ......................................................................................... 362
3.6. Sterilization .......................................................................................... 362
4. Plantlet growth, production costs and sales price ....................................... 362
4.1 Calla lily plantlet growth....................................................................... 362
4.2. China fir plantlet growth ...................................................................... 365
4.3. Percent survival during acclimatization ex vitro.................................. 366
4.4. Production cost of calla lily plantlets: A case study ............................ 367
4.4.1. Production cost per acclimatized plantlet ..................................... 368
4.4.2. Cost, labour and electricity consumption for multiplication
or rooting................................................................................................ 368
4.4.3. Sales price of in vitro and ex vitro acclimatized plantlets ............ 370
5. Conclusions ................................................................................................. 370
Acknowledgement........................................................................................... 370
References ....................................................................................................... 370
Intelligent inverse analysis for temperature distribution in a plant
culture vessel ..................................................................................................... 373
H. Murase, T. Okayama, and Suroso .................................................................. 373
1. Introduction ................................................................................................. 373
2. Theoretical backgrounds ............................................................................. 375
3. Methodology ............................................................................................... 378
3.1. Finite element neural network inverse technique algorithm................ 378
3.2. Finite element formulation ................................................................... 379
3.3. Finite element model............................................................................ 380
3.4. Neural network structure...................................................................... 381
3.5. Neural network training ....................................................................... 381
3.6. Optimization of temperature distribution inside the culture vessel ..... 382
3.6.1. Genetic algorithm flowchart ......................................................... 382
3.6.2. Objective function ......................................................................... 383
3.6.3. Genetic reproduction ..................................................................... 383
3.7. Temperature distribution measurement................................................ 386
3.7.1. Equipment development for temperature distribution
measurement............................................................................................ 386
3.7.2. Temperature distribution data ....................................................... 388
4. Example of solution .................................................................................... 388
4.1. Coefficient of convective heat transfer ................................................ 388
4.2. Verification of the calculated coefficient of convective heat transfer . 390
4.3. Optimum values of air velocity and bottom temperature .................... 391
References ....................................................................................................... 394
PART 5.................................................................................................................... 395
PHYSICAL ASPECTS OF PLANT TISSUE ENGINEERING............................. 395
Electrical control of plant morphogenesis ...................................................... 397
Cogălniceanu Gina Carmen ................................................................................ 397
1. Introduction ................................................................................................. 397
2. Endogenous electric currents as control mechanisms in plant development 397
3. Electrostimulation of in vitro plant development ....................................... 400
9
Table of Contents
4. High-voltage, short-duration electric pulses interaction with in vitro
systems.............................................................................................................
4.1. Effects of electric pulses treatment on plant protoplasts .....................
4.2. Effects of electric pulses treatment on tissue fragments or entire
plantlets........................................................................................................
5. Potential applications of the electric manipulation in plant biotechnology
References .......................................................................................................
The uses of ultrasound in plant tissue culture ................................................
Victor Gaba, K. Kathiravan, S. Amutha, Sima Singer, Xia Xiaodi and
G. Ananthakrishnan ............................................................................................
1. Introduction .................................................................................................
2. The generation of ultrasound ......................................................................
3. Mechanisms of action of ultrasound ...........................................................
4. Sonication-assisted DNA transformation....................................................
5. Sonication-assisted Agrobacterium-mediated transformation ....................
6. Stimulation of regeneration by sonication ..................................................
7. Summary of transformation and morphogenic responses to ultrasound.....
8. Fractionation of somatic embryos...............................................................
9. Secondary product synthesis .......................................................................
10. Ultrasound and control of micro-organisms .............................................
11. Conclusions ...............................................................................................
References .......................................................................................................
Acoustic characteristics of plant leaves using ultrasonic transmission
waves....................................................................................................................
Mikio Fukuhara, S. Dutta Gupta and Limi Okushima ........................................
1. Introduction .................................................................................................
2. Theoretical considerations and system description.....................................
3. Case studies on possible ultrasonic diagnosis of plant leaves ....................
3.1. Ultrasonic testing of tea leaves for plant maturity ...............................
3.1.1. Wave velocity and dynamic modulus for leaf tissue development
3.1.2. Dynamic viscosity and imaginary parts in complex waves ..........
3.2. Ultrasonic diagnosis of rice leaves.......................................................
3.3. Acoustic characteristics of in vitro regenerated leaves of gladiolus....
4. Conclusions .................................................................................................
Acknowledgement...........................................................................................
References .......................................................................................................
Physical and engineering perspectives of in vitro plant cryopreservation...
Erica E. Benson, Jason Johnston, Jayanthi Muthusamy and Keith Harding ......
1. Introduction .................................................................................................
2. The properties of liquid nitrogen and cryosafety ........................................
3. Physics of ice...............................................................................................
3.1. Water’s liquid and ice morphologies ...................................................
3.1.1. Making snowflakes: a multiplicity of ice families........................
4. Cryoprotection, cryodestruction and cryopreservation...............................
4.1. Physical perspectives of ultra rapid and droplet freezing ....................
10
403
404
406
410
411
417
417
417
418
419
420
420
421
422
423
423
423
424
424
424
427
427
427
428
430
430
431
432
434
435
438
438
438
441
441
441
442
443
444
445
447
448
Table of Contents
4.2. Controlled rate or slow cooling............................................................
4.3. Vitrification ..........................................................................................
5. Cryoengineering: technology and equipment .............................................
5.1. Cryoengineering for cryogenic storage................................................
5.1.1. Controlled rate freezers .................................................................
5.1.2. Cryogenic storage and shipment ...................................................
5.1.3. Sample safety, security and identification ....................................
6. Cryomicroscopy ..........................................................................................
6.1. Nuclear imaging in cryogenic systems ................................................
7. Thermal analysis .........................................................................................
7.1. Principles and applications...................................................................
7.1.1. DSC and the optimisation of cryopreservation protocols .............
7.1.2. A DSC study comparing cryopreserved tropical and temperate
plant germplasm ......................................................................................
7.1.2.1. Using thermal analysis to optimise cryoprotective strategies....
8. Cryoengineering futures..............................................................................
References .......................................................................................................
INDEX.....................................................................................................................
11
450
451
451
451
452
455
456
456
458
459
460
462
463
468
470
473
474
477
This page intentionally blank
PART 1
MACHINE VISION
EVALUATION OF PHOTOSYNTHETIC CAPACITY IN
MICROPROPAGATED PLANTS BY IMAGE ANALYSIS
YASUOMI IBARAKI
Department of Biological Science, Yamaguchi University, Yoshida 16771, Yamaguchi-shi, Yamaguchi 753-8515, Japan – Fax: +81-83-933-5864
Email: [email protected]
1. Introduction
In micropropagation, in vitro environmental conditions (i.e., environmental conditions
surrounding plantlets within culture vessels such as light conditions, temperature, and
gaseous composition), have an important role in plantlet growth. Normally, in vitro
environmental conditions cannot be controlled directly; instead, they are largely
determined by regulated culture conditions outside the vessel. Therefore, culture
conditions should be optimized for plantlet growth. It is necessary for optimization of
culture conditions to understand relationships between culture conditions and in vitro
plant growth, physiological state, or both. In vitro environmental conditions may change
with plantlet growth during culture because the plantlet itself affects them. Therefore,
non-destructive evaluation of the growth of micropropagated plantlets and their
physiological state without disturbing the in vitro environmental conditions is desirable
for investigating these relationships and considering their dynamics.
Recent studies revealed that in vitro cultured chlorophyllous plantlets had
photosynthetic ability but their net photosynthetic rates were restricted by
environmental conditions [1]. The photosynthetic properties of plantlets in vitro depend
on culture conditions, including light intensity [2], the degree of air exchange between a
vessel and the surrounding air [3], and the sugar content in the medium [4].
Photoautotrophic micropropagation which is micropropagation with no sugar added to
the medium has many advantages, especially in plantlet quality [1]. For successful
photoautotrophic micropropagation, in vitro environmental conditions should be
properly controlled to enhance photosynthesis of the plantlets by manipulation of
culture conditions. Successful photoautotrophic micropropagation also requires
knowledge of when cultures should transit from photomixotrophic into
photoautotrophic [1]. An understanding of changes in photosynthetic properties of
cultured plantlets during the culture period is essential to optimize culture conditions for
photoautotrophic culture to obtain high-quality plantlets.
It is difficult to evaluate photosynthetic properties of plantlets non-destructively. Carbon
dioxide gas exchange rates of plantlets in vitro can be estimated in situ by measurements
of the concentration of CO2 inside and outside the culture vessel, the degree of air
15
S. Dutta Gupta and Y. Ibaraki (eds.), Plant Tissue Culture Engineering, 15–29.
© 2008 Springer.
www.taq.ir
Y. Ibaraki
exchange between the vessel and the surrounding air, and the head space volume in the
vessel [5]. However, the estimated gas exchange rates are the rates per all plantlets
within the vessel, and they should be converted to the rates per unit leaf area or unit dry
weight for analysis of the photosynthetic properties. This requires estimation of leaf
area or dry weight of plantlets in the vessel. In addition, it should be noted that the
environmental conditions could be non-uniform in a culture vessel even under
controlled culture conditions. In culture vessels, air movement is limited, and as a
result, there may be gradients in humidity and/or CO2 concentration within the vessels.
In addition, vertical light intensity distribution exists in slender vessels like test tubes
[6]. This might cause variations in the in vitro microenvironment around the cultured
plants and consequently cause variations in photosynthetic capacity. This variation may
affect uniformity in plantlet quality, especially when propagating by cuttings, such as
for potato nodal cutting cultures. An understanding of variations in photosynthetic
properties within cultured plantlets may be helpful for obtaining uniform-quality
plantlets.
Chlorophyll fluorescence has been a useful tool for photosynthetic research. In
recent years, the value of this tool in plant physiology has been greatly increased by the
availability of suitable instrumentation and an increased understanding of the processes
that regulate fluorescence yield [7]. It has enabled analysis of the photosynthetic
properties of plant leaves, especially characteristics related to the photochemical
efficiency of photosystem II. As chlorophyll fluorescence analysis is based on
photometry, i.e., measurement of light intensity, it is a promising means of nondestructive estimation of photosynthetic capacity.
In this chapter, the methods for non-destructive evaluation of photosynthetic
capacity are introduced, focusing on imaging of chlorophyll fluorescence. First, the
principle of photosynthetic analysis based on chlorophyll fluorescence will be outlined,
and the feasibility of imaging the chlorophyll fluorescence parameters for
micropropagated plants from outside the culture vessels will be discussed. Other
promising indices based on spectral reflectance for imaging the photosynthetic capacity
of micropropagated plants will be also discussed. In addition, estimation methods for
light intensity distribution inside culture vessels will be introduced in consideration of
its influence on the photosynthetic properties of cultured plants.
2. Basics of chlorophyll fluorescence
Chlorophyll absorbs photons for use in the photochemical reaction of photosynthesis.
Excited chlorophyll can re-emit a photon and return to its ground state, and this
fluorescence is called chlorophyll fluorescence. Occasionally, it is also referred to as
chlorophyll a fluorescence, since it is due to chlorophyll a. The analysis of chlorophyll
fluorescence provides a powerful probe of the functioning of the intact photosynthetic
system [8]. It especially enables us to obtain information on the functioning of
photosystem II (PSII), since at room temperature chlorophyll fluorescence is
predominantly derived from PSII [9]. Methods to analyze photosynthetic properties of
leaves using chlorophyll fluorescence include a method using a saturating light pulse
and another method based on induction kinetics (the Kautsky curve [10]). Here, the
16
www.taq.ir
Evaluation of photosynthetic capacity in micropropagated plants by image analysis
former method, in which fluorescence is measured while varying PSII photochemical
efficiency using a saturating light pulse, is more fully explained.
After dark adaptation treatment, the yield, ĭF of fluorescence excited by very weak
irradiance is expressed by the following equation:
)F
kF
k F k D kT k P
(1)
Where kF, kD, kT, and kP are rate constants for fluorescence, thermal dissipation, energy
transfer to PSI and PSII photochemistry (electron transport), respectively.
As the portion of energy transfer is very small, kT can be neglected in the above
equation [7]. This fluorescence, which occurs when the primary electron acceptor, QA,
is fully oxidized due to excitation by weak light just after dark adaptation, is referred to
as Fo. Then, irradiation by a saturating light pulse (of very high intensity) leads to full
reduction of QA (sometimes the condition is referred to as “closed”). The fluorescent
yield, ĭFm, of maximum fluorescence Fm, determined under the saturating light pulse,
is expressed by the following equation:
) Fm
kF
k F k D kT
(2)
From Fo and Fm, the maximum quantum yield of PSII, Fv/Fm, is estimated using the
following equation:
Fv/Fm
Fm Fo
Fm

½
½
kF
kF
kF
®
¾/ ®
¾
¯ k F k D kT k F k D kT k P ¿ ¯ k F k D kT ¿

k F k D kT ½
®1 ¾
¯ k F k D kT k P ¿
kP
k F k D kT k P
(3)
Fv/Fm is a measure of photoinhibition and has been used for photosynthetic capacity
evaluation in photosynthetic research (e.g., [11]) and cultivar screening (e.g., [12]).
Under light conditions without dark adaptation (hereafter, the light is referred to as
actinic light to distinguish from the light for fluorescent measurements), the actual
quantum yield of PSII, ĭPSII, can be also estimated using the following equation:
17
www.taq.ir
Y. Ibaraki
ĭPSII
'F/Fm'
Fm' F
Fm'
(4)
Where F is the fluorescence excited by the measuring light under the actinic light, and
Fm’ is the fluorescence excited by the measuring light while irradiating with the
saturating light pulse (that is, when QA is fully closed) under the actinic light. As for the
other parameters, photochemical quenching, qp, which shows the extent to which
ĭPSII is restricted by photochemical capacity at PSII, and indices of non-photochemical
quenching, qN and NPQ, which are related to heat dissipation, can be derived by
fluorescence measurement using a saturating light pulse. Also, the linear electron
transport rate, ETR, can be estimated if the number of photons absorbed is known [13].
These parameters were reviewed by Maxwell and Johnson in detail [14]. The
chlorophyll fluorescence parameters can be measured by a pulse amplitude modulation
(PAM) fluorometer. In this fluorometer, the excitation light (pulsed light of low
intensity; hereafter, measuring pulse) used to measure chlorophyll fluorescence is
separately applied to the actinic light, which drives the photosynthetic light reaction
[15]. Due to the selective pulse-amplification system, only fluorescence excited by the
measuring pulse is recorded in the presence of the actinic light [15]. Although in some
cases the parameters can be obtained non-destructively with PAM fluorometer, there are
some limitations in the measurements, for example due to the short distance (10-15
mm) between the sensor probe of the fluorometer and the leaf surface.
3. Imaging of chlorophyll fluorescence for micropropagated plants
3.1. CHLOROPHYLL FLUORESCENCE IN IN VITRO CULTURED PLANTS
In research on micropropagation, the chlorophyll fluorescence parameter Fv/Fm has
been used to evaluate photosynthetic capacity, though applications are limited to a few
studies. The nutrient composition of the medium affects Fv/Fm of in vitro cultured
Pinus radiata [16]. Ex vitro transfer for acclimatization causes a decrease in Fv/Fm of
plantlets and the degree of the reduction in Fv/Fm depended on culture conditions
[17,18]. In general, plants grown under low light intensity are more sensitive to
photoinhibition caused by high light intensity [19]. Therefore, Fv/Fm of
micropropagated plantlets may be subject to change according to culture conditions.
18
www.taq.ir
Table 1. Fv/Fm of potato plantlets of different sucrose content treatments (Exp.1).
Reproduced from Ibaraki, Y. and Matsumura, K. (2004) [20].
Fv/Fm
Average
CV*
30 g/L
0.795 b**
0.032 ab**
10 g/L
0.750 c
0.055 a
0 g/L
0.818 a
0.020 b
* Coefficient of variation in a single plantlet, ** Different letters within row show significant differences by
Tukey multiple range test at 1% level
Table 2. Fv/Fm of potato plantlets of different sucrose content treatments (Exp.2).
Fv/Fm
Average
CV*
30 g/L
0.77 a**
0.032 b**
0 g/L
0.72 b
0.115 a
* Coefficient of variation in a single plantlet, ** Different letters within row show significant differences by
Tukey multiple range test at 1% level.
To investigate sensitivity of Fv/Fm to culture conditions, two experiments were
conducted to determine Fv/Fm for potato plantlets cultured under various environmental
conditions [20]. In one experiment, potato nodal cuttings were transplanted into glass
tubes containing MS medium [21] with different contents of sucrose (30 g/L, 10 g/L,
and 0 g/L). In the case of the sugar-free treatment, a hydrophobic Fluoropore®
membrane filter (Milliseal®, Millipore®) was attached to the plastic cap of the glass tube
to enhance gas exchange for photoautotrophic growth. In another experiment, Fv/Fm
values of plantlets cultured in medium with 30 g/L sucrose or in sugar-free medium were
compared under conditions where gas exchange was suppressed using normal plastic
caps for both treatments. At the end of culturing (35d and 40d after transplanting for
experiment 1 and experiment 2, respectively), plantlets were transferred ex vitro, and
Fv/Fm was measured randomly for all measurable leaves of the plantlets using a PAM
fluorometer (MINI-PAM, Walz, Germany) after a 60 min dark adaptation treatment. For
each treatment, 8 plantlets were tested. Average Fv/Fm values were affected by culture
conditions (Tables 1 and 2). Without promoting gas exchange of culture vessels, Fv/Fm
values of plantlets cultured in sugar-free medium were lower than for plantlets in 30 g/L
sucrose treatment, which is a conventional medium formulation. In contrast, plantlets
cultured with sugar-free medium in culture vessels promoting gas exchange showed
19
www.taq.ir
Y. Ibaraki
higher Fv/Fm than plantlets cultured in medium containing 30 g/L sucrose, indicating a
higher photochemical efficiency. Combined effects of enhanced gas exchange and
omission of sucrose from the medium might improve photosynthetic capacity. In
comparisons between sucrose-containing treatments (experiment 1), plantlets of the 10
g/L treatment showed a lower Fv/Fm than plantlets of the 30 g/L treatment, and also
suppressed growth. Variations in Fv/Fm values were observed among the plantlets and
the distribution patterns in a plantlet changed slightly with sucrose content (Figures 1
and 2).
Figure 1. Fv/Fm distribution in potato plantlets cultured in MS medium contained 30 g/L,
10 g/L, or 0 g/L sucrose for 35 d (Exp. 1). Reproduced from Ibaraki, Y. and Matsumura, K.
(2004) [20]. In sugar-free treatment, gas exchange was promoted by using the cap attached
a hydrophobic Fluoropore (R) membrane filter. Lower 3 leaves, upper 3 leaves, and other
leaves were classified into lower, upper, and middle in leaf position, respectively. Bar, SE.
Different letters on graph lines show significant differences among leaf positions by Tukey
multiple range test at 1% level.
These results suggest that Fv/Fm may change according to culture conditions, and that
analysis of Fv/Fm for evaluation of photosynthetic capacity of cultured plantlets is
effective for optimization of culture conditions.
Although Fv/Fm measurement is simple with the PAM fluorometer, there are some
difficulties in measurements of plantlets within the culture vessel through the culture
vessel wall. The measurement requires fixing the short distance between the sensor
probe of the fluorometer and the leaf surface. This is a difficult requirement for plantlet
leaves in a culture vessel. In addition, measurements for small leaves of plantlets with
the fluorometer were subject to errors [20]. Non-destructive methods suited for
micropropagated plants are desirable.
20
www.taq.ir
Figure 2. Fv/Fm distribution in potato plantlets cultured in MS medium contained 30 g/L or
0 g/L sucrose for 40 d (Exp. 2). Lower 3 leaves, upper 3 leaves, and other leaves were
classified into lower, upper, and middle in leaf position, respectively. Bar, SE. Different
letters on graph lines show significant differences among leaf positions by Tukey multiple
range test at 1% level.
In a few studies, the chlorophyll fluorescence parameter 'F/Fm’, determined under
actinic light by PAM fluorometer, has been used in micropropagation research. Since
'F/Fm’ depends on the level of light irradiating a leaf, and it is difficult to know the
exact irradiation level, careful consideration is required to determine photosynthetic
properties from values of 'F/Fm’. If the same light intensity were set for all plantlets
tested, or if the light intensity distribution could be determined in culture vessels,
'F/Fm’ would offer information on plantlet photosynthetic capacity.
3.2. IMAGING OF CHLOROPHYLL FLUORESCENCE
Imaging of chlorophyll fluorescence was first reported by Omasa et al. [22]. In this
study, the kinetics of chlorophyll fluorescence was analyzed using fluorescent images.
For cultured callus and plantlets of Daucus carota, images of chlorophyll fluorescence
induction were also used to analyze the development of photosynthetic apparatus [23].
Although several studies on chlorophyll fluorescence imaging had been reported, these
primary studies required empirical calibration of the fluorescent signal using other
methods, such as gas exchange, when the fluorescence images were converted to images
of photosynthesis [24]. Recently, several reports showed the possibility of imaging
chlorophyll fluorescence parameters based on a saturating light pulse method in order to
obtain an image of photochemical efficiency over a leaf. Genty and Meyer [24]
developed a method to construct the topography of the photochemical quantum yield of
PSII and showed the effectiveness of the method by mapping the heterogeneous
21
www.taq.ir
Y. Ibaraki
distribution of photosynthetic activity after treatment with an herbicide, with abscisic
acid, or during the course of induction of photosynthesis. Oscillations in photosynthesis
initiated by a transient decrease in light intensity could be imaged over the leaf [25].
The sink-source transition of developing tobacco leaves was analyzed using images to
evaluate electron transport rates [26]. Oxborough and Baker [7] proposed a method to
image not only photochemical quantum yield but also non-photochemical quenching,
assumed to correspond mainly to heat dissipation. In addition, Oxborough and Baker
[27] developed a system to image Fo and consequently obtain an Fv/Fm image using a
fluorescence microscope and a cooled charge coupled device (CCD) camera.
Chlorophyll fluorescence parameters can be imaged by considering the following
points: 1) to distinguish between fluorescence and reflection by use of optical filters,
and 2) to measure fluorescent quantum yield. Basic device arrangements for imaging of
chlorophyll fluorescence include a light source for excitation of fluorescence, a camera,
and optical filters for controlling excitation light intensity and separating reflected light
and fluorescence. Normally, fluorescent intensity can be imaged as the grey level in
each pixel by the camera. Therefore, it is necessary to convert fluorescent intensity into
fluorescent yield to construct images mapping chlorophyll fluorescence parameters. If
the irradiance distribution on a leaf were determined exactly, it would be possible to
convert the fluorescent intensity to fluorescent yield. Actually, the conversion is done
by controlling exposure time according to excitation light intensity [24], by imaging a
fluorescent standard at the same time [25], or by imaging a reference leaf at the same
time [20]. Recently, a PAM-based fluorescence imaging system (IMAGING-PAM,
Walz, Germany) has been developed, which is now available. Although there have been
few studies using the system to date, it is promising for non-destructive evaluation of
plant photosynthetic properties.
For selection of cameras to image fluorescence, some considerations are required. In
Fv/Fm measurements, Fo is not intense because it is excited by very low irradiance, so
highly sensitive cameras such as expensive cooled CCD cameras are needed. Although
low-cost CCD cameras with high sensitivity have become available recently, the images
acquired by most have reduced numbers of distinct grey levels. It is necessary to discuss
whether the number of distinct grey levels in an image is sufficient for calculations used
to derive chlorophyll parameters. In addition, gamma and auto-gain features of cameras
should be carefully treated because they affect the relationship between light intensity
and the pixel grey level value. The relationship between light intensity and the pixel
grey level value in the image should be calibrated using a fluorescent or grey standard.
3.3. IMAGING OF CHLOROPHYLL FLUORESCENCE IN MICROPROPAGATED
PLANTS
A system for imaging chlorophyll fluorescence of leaves of Solanum tuberosum plantlet
from the outside of culture vessels and for estimating the fluorescence parameter Fv/Fm
was developed [20].
22
www.taq.ir
Figure 3. Schematic layout of a chlorophyll fluorescence imaging system. Reproduced from
Ibaraki, Y. and Matsumura, K. (2004) [20].
Figure 3 shows the schematic layout of the system. The plantlets in glass test tubes were
illuminated by a halogen lamp with a light fiber (HL-150, Hoya-Schott, Japan), and the
light intensity for fluorescence excitation was controlled by neutral density filters (S-7350-3,-13, Suruga, Japan). Fluorescence was imaged by a highly sensitive
monochromatic CCD camera (WAT-120N, Watec, Japan) with long path filters. Fv/Fm
was estimated from the Fo image, which was a fluorescent image acquired under low
intensity illumination (0.15 Pmol m-2 s-1) after a 60 min dark adaptation treatment, and
the Fm image, which was then acquired under high intensity illumination (2500 Pmol
m-2 s-1). A detached Epipremnum aureum leaf, with a predetermined Fv/Fm, was imaged
together as a reference leaf, and used to calibrate the fluorescence image. The Fv/Fm
image (IFvFm) was constructed as a pixel-by-pixel calculation of the Fo image (IFo) and
the Fm image (IFm) by the following equation:
I FvFm
I Fm kI Fo (5)
I Fm
Where, k is a coefficient that is used to convert fluorescent intensity into fluorescent
yield and was determined so as to fit the estimated Fv/Fm of the reference leaf by
equation 1 to the Fv/Fm measured before imaging by the fluorometer (MINI-PAM,
Walz, Germany).
Figure 4 shows examples of chlorophyll fluorescence images, and Fv/Fm images
derived from them, of potato plantlets using the system. For a few leaves of the plantlets,
Fv/Fm could be imaged at the same time. Therefore, using images acquired repeatedly
after dark-adaptation treatment, the Fv/Fm distribution in an individual plantlet could be
determined. Changes in Fv/Fm of an individual leaf over a culture period could also be
detected using the system. Figure 5 shows the changes in Fv/Fm of the 5th leaf
determined by the fluorescence imaging system developed. The leaf just expanded (14 d
after transplanting) showed a lower Fv/Fm (<0.8). Then, Fv/Fm increased and decreased
23
www.taq.ir
Y. Ibaraki
again after a peak at 14 d after leaf expansion. This was a reasonable pattern in Fv/Fm
changes, since a decline of Fv/Fm was reported in young leaves and older leaves [28].
The system enabled gathering of information on photosynthetic capacity of cultured
plantlets from the outside of culture vessels non-destructively. The system should be
useful for optimizing culture conditions.
Figure 4. An example of Fv/Fm images constructed from Fo image and Fm image acquired
by the chlorophyll fluorescence imaging system. Reproduced from Ibaraki, Y. and
Matsumura, K. (2004)[20]. A circle in Fo image is an area to be used as the reference in
the potato leaf.
Figure 5. Changes in Fv/Fm of the 5th leaf of a potato plantlet at intervals of 7d.
Reproduced from Ibaraki, Y. and Matsumura, K. (2004) [20].
24
www.taq.ir
4. Techniques for image-analysis-based evaluation of photosynthetic capacity
Spectral reflectance has been used to obtain plant growth information, especially in the
research area of remote sensing. As spectral reflectance measurements are based on
photometry, they have potential for non-destructive evaluation of plant growth and
physiological state. The normalized difference vegetation index (NDVI), which can be
calculated by reflectance at red and near infrared (NIR) wavelengths, has been widely
used for monitoring, analyzing, and mapping temporal and spatial distributions of
physiological and biophysical characteristics of vegetation [29]. It is applied not to an
individual leaf, but to a plant canopy or wider area such as a forest, and is used mainly
for quantification of vegetation, such as estimation of specific leaf area and evaluation
of plant activity. The chlorophyll content of leaves can be estimated using the ratio of
reflectance at 675 nm and 700 nm [30] or at 695 nm and 760 nm [31]. Although these
indices are not a direct measure of photosynthetic capacity, they would be usable if
empirical relationships between indices and photosynthetic capacity estimated by other
methods could be determined.
Recently, the photochemical reflectance index (PRI) was proposed for estimation of
photosynthetic radiation use efficiency [32]. This index is derived from reflectance at
531 nm and 570 nm, and is a measure of the degree of the photo-protective xanthophyll
cycle pigment, zeaxanthin. The xanthophyll cycle, where the carotenoid pigment
violaxanthin is converted to antheraxanthin and zeaxanthin via de-epoxidase reactions
[33], is related to heat dissipation. The PRI is highly correlated with quantum yield of
PSII determined by chlorophyll fluorescence for 20 species representing three
functional types of plants [32]. Stylinski et al. [34] also reported a strong correlation of
PRI to the chlorophyll fluorescence parameter 'F/Fm’ across species and seasons. As
described previously, light use efficiency can vary with incident light intensity.
Although several limitations still remain, the use of PRI is promising for evaluating
photosynthetic capacity by a machine vision system.
Figure 6. A concept illustration of a PRI imaging system.
25
www.taq.ir
Y. Ibaraki
Figure 6 shows a concept for a hypothetical PRI imaging system. In measurement of
PRI, reflectance images should be acquired at two different wavelengths (531 and
570 nm). For this purpose, each image is taken with a grey standard by the CCD camera
with a narrow-band-pass filter for the respective wavelength. The grey standard has
nearly constant reflectance over the visible spectrum and is used to determine relative
reflectance from light intensity. Configurations of the light source, the object (the
culture vessel), and the camera should be carefully determined to collect the diffuse
reflectance while reducing total internal reflection. Carter et al. [35] proposed a system
using the same concept for reflectance imaging for early detection of plant stress.
5. Estimation of light distribution inside culture vessels
5.1. UNDERSTANDING LIGHT DISTRIBUTION IN CULTURE VESSELS
One of the most important factors for photosynthesis of cultured plantlets during
micropropagation is the light environment, especially light intensity. High light
intensity with sufficient CO2 supply can enhance plantlet growth [36] and has the
potential to facilitate acclimatization. From the viewpoint of photosynthesis, light
intensity should be evaluated by photosynthetic photon flux density (PPFD) on the
plantlet. However, since PPFD on plantlets is difficult to measure in a small culture
vessel, it is usually represented by the value determined outside the vessel. PPFD on
plantlets depends on the material and shape of culture vessels, the position of the vessel
on the culture shelf, the position of the light sources, the optical characteristics of the
shelf, etc [37]. It should be noted that PPFD in culture vessels with a closure, even with
a high light transmissivity, was significantly lower than that on the empty shelf [38].
Moreover, when long culture vessels such as test tubes are used, light intensity can
differ greatly between the top and bottom of the vessel. Non-uniform light distribution
in a culture vessel may be responsible for differences in photosynthetic capacity and/or
growth among leaves in the plantlet. As a result, this may lead to variations in plantlet
quality in the case of a nodal cutting culture such as potato [6]. The estimation of light
intensity distribution inside culture vessels is important for understanding the
relationship between culture conditions and cultured plantlet growth properly. The use
of information on light distribution in a culture vessel with information on
photosynthetic capacity determined non-destructively would be helpful for optimization
of culture conditions.
5.2. ESTIMATION OF LIGHT DISTRIBUTION WITHIN CULTURE VESSELS
A recently developed sensor film for measuring integrated solar radiation (Optleaf®),
Taisei Chemical Co. Ltd., Japan) potentially offers a simple technique to estimate light
intensity distribution. It has been used previously to estimate light intensity distribution
in plant canopy (e.g., [39]). Here, the method [6] to estimate light intensity distribution
inside a small culture vessel using the small piece of the sensor film is introduced.
This method enabled us to estimate light intensity distribution inside a culture vessel
using a plantlet model whose leaves were constructed from sensor film. A plantlet
26
www.taq.ir
model simulating a potato plantlet consisted of 8 model leaves fabricated from sensor
films (Optleaf R-2D, Taisei Chemical Co. Ltd., Japan) for measuring integrated solar
radiation and a wire stem. A leaf-shaped piece of sensor film (dimensions 10 mm x 7
mm) was attached to an identically shaped piece of white paper and fixed to the wire
stem at an angle of 30q. Each leaf was set at vertical intervals of 12 mm and at a
horizontal angular interval of 120q. The total height of the plantlet model was 135 mm.
A glass tube (25 mm x 150 mm) with a transparent plastic cap was used as the culture
vessel. The sensor film was a cellulose acetate film coloured by azo dyes. Integrated
radiation was estimated based on the degree of fading of the sensor film, which was
quantified by measuring transmittance at 470 nm with a photometer (THS-470, Taisei
Chemical Co. Ltd., Japan). Normally, measurements are performed while the film is set
to a film mount (accessory of the photometer), but the model leaf was so small that the
film mount could not be used. Therefore, the model leaf was set on 100% transmittance
adjustment film (accessory of the photometer). The linear model determined previously
could be used to correct the transmittance of model leaves. The sensor film absorbance
was calculated from the sensor film transmittance and the ratio of the sensor film
absorbance after exposure to that before exposure (film fading ratio) was determined.
Integrated radiation was determined from the film fading ratio using a calibration curve
provided by the film manufacturer (Taisei Chemical Co. Ltd., Japan).
Culture vessels with plantlet models were set on the shelf being surrounded with
vessels containing potato plantlets in a temperature-controlled growth chamber at 24qC.
Fluorescent tubes illuminated the growth chamber from the top (downward lighting)
and the distance between the surface of fluorescent tubes and the top of vessels was 10
mm. In downward lighting condition, PPFD decreased toward the bottom of the vessel
and was reduced to 50% and 30% of the maximum at the middle and the lower leaves,
respectively. As compared with the PPFD measured with the photon sensor at the same
position as each leaf position outside the vessel without the surrounding vessels, the
steeper decline in PPFD inside the vessel could be observed. This might be due to
interception of light by upper leaves and the surrounding vessels. PPFD distribution
pattern inside the vessel can differ from that outside the vessel.
The results demonstrate that the use of sensor film plantlet models enables light
intensity distribution inside a small culture vessel to be estimated, which was previously
assumed to be too difficult to measure. This method could be applied to the
determination of light intensity distribution patterns inside various types of culture
vessels and under various lighting conditions, and thus would be of value in the
optimization of culture conditions.
6. Concluding remarks
Non-destructive measurements of photosynthetic properties of plants in culture vessels
are useful for understanding relationships between culture conditions and photosynthetic
capacity, offering data on changes in physiological state of the plants during culturing
without disturbing the in vitro microenvironment. Chlorophyll fluorescence has potential
for non-destructive evaluation of leaf photosynthetic properties because the
measurement can be conducted based on photometry. Parameters derived from
chlorophyll fluorescence measurements relate to the functioning of PSII, including the
27
www.taq.ir
Y. Ibaraki
maximum quantum yield. Image analysis yielding these parameters is promising for
non-destructive evaluation of photosynthetic capacity of micropropagated plants.
References
[1] Kubota, C. (2001) Concepts and background of photoautotrophic micropropagation. In: Morohoshi, N.
and Komamine, A. (Eds.) Molecular Breeding of Woody Plants. Elsevier Science B.V., Amsterdam; pp.
325-334.
[2] Dubé, S.L. and Vidaver, W. (1992) Photosynthetic competence of plantlets grown in vitro. An automated
system for measurement of photosynthesis in vitro. Physiol. Plant 84: 409-416.
[3] Kubota, C. and Kozai, T. (1992) Growth and net photosynthetic rate of Solanum tuberosum in vitro under
forced and natural ventilation. Hort. Sci. 27: 1312-1314.
[4] Capellades, M.; Lemeur, R. and Debergh, P. (1990) Effects of sucrose on starch accumulation and rate of
photosynthesis in Rosa cultured in vitro. Plant Cell Tissue Org. Cult. 25: 21-26.
[5] Desjardins, Y.; Hdider, C. and de Riek, J. (1995) Carbon nutrition in vitro – regulation and manipulation
of carbon assimilation in micropropagated systems. In: Aitken-Christie, J.; Kozai, T. And Smith, M.A.L.
(Eds.) Automation and Environmental Control in Plant Tissue Cultures. Kluwer Academic Publishers,
Dordrecht, The Netherlands; pp. 441-471.
[6] Ibaraki, Y. and Nozaki, Y. (2004) Estimation of light intensity distribution in a culture vessel. Plant Cell
Tissue Org. Cult. (in press).
[7] Oxborough, K. and Baker, N.R. (1997) Resolving chlorophyll a fluorescence images of photosynthetic
efficiency into photochemical and non-photochemical components-calculation of qp and Fv’/Fm’ without
measuring Fo’. Photosynth. Res. 54: 135-142.
[8] Jones, H.G. (1990) Plants and microclimate. Cambridge University Press, New York.
[9] Lichtenthaler, H.K.; Lang, M.; Sowinska, M.; Heisel, F. and Miehe, J.A. (1996) Detection of vegetation
stress via a new high resolution fluorescence imaging system. J. Plant Physiol. 148: 599-612.
[10] Lichtenthaler, H.K.; Buschman, C.; Rinderle, U. and Schmuck, G. (1986) Application of chlorophyll
fluorescence in eco-physiology. Radiat. Environ. Biophy. 25: 297.
[11] Morecroft, M.D.; Stokes, V.J. and Morison, J.I.L. (2003) Seasonal changes in the photosynthetic
capacity of canopy oak (Quercus robur) leaves: the impact of slow development on annual carbon
uptake. Int. J. Biometeorol. 47: 221-226.
[12] Fracheboud, Y.; Haldimann, P.; Leipner, J. and Stamp, P. (1999) Chlorophyll fluorescence as a selection
tool for cold tolerance of photosynthesis in maize (Zea mays L.). J. Exp. Bot. 50: 1533-1540.
[13] Genty, B.; Briantais, J.M. and Baker, N.R. (1989) The relationship between the quantum yield of
photosynthetic electron transport and quenching of chlorophyll fluorescence. Biochemica Biophysica
Acta 990: 87-92.
[14] Maxwell, K. and Johnson, G.N. (2000) Chlorophyll fluorescence – a practical guide. J. Exp. Bot. 51:
659-668.
[15] Lichtenthaler, H.K. and Rinderle, U. (1988) The role of chlorophyll fluorescence in the detection of
stress conditions in plants. CRC Critical Reviews in Analytical Chemistry 19: S29-S85.
[16] Aitken-Christie, J.; Davies, H.E.; Kubota, C. and Fujiwara, K. (1992) Effect of nutrient media
composition on sugar-free growth and chlorophyll fluorescence of Pinus radiata shoots in vitro. Acta
Hort. 319: 125-128.
[17] Hofman, P.; Haisel, D.; Komenda, J.; Vágner, M.; Tichá, I.; Schäfer, C. and ýapková, V. (2002) Impact
of in vitro cultivation conditions on stress responses and on changes in thylakoid membrane proteins and
pigments of tobacco during ex vitro acclimation. Biol. Plant. 45: 189-195.
[18] Serret, M.D.; Trillas, M.I. and Araus, J.L. (2001) The effect of in vitro culture conditions on the pattern
of photoinhibition during acclimation of gardenia plantlets to ex vitro conditions. Photosynthetica 39: 6773.
[19] Kato, M.C.; Hikosaka, K. and Hirose, T. (2002) Leaf discs floated on water are different from intact
leaves in photosynthesis and photoinhibition. Photosynth. Res. 72: 65-70.
[20] Ibaraki, Y and Matsumura, K (2004) Non-destructive evaluation of the photosynthetic capacity of PSII in
micropropagated plants. J. Agric. Meteorol. 60 (in press).
[21] Murashige, T. and Skoog, F. (1962) A revised medium for rapid growth and bioassays with tobacco
tissue cultures. Physiol. Plant 15: 473-497.
28
www.taq.ir
[22] Omasa, K.; Shimazaki, K.I.; Aiga, I.; Larcher, W. and Onoe, M. (1987) Image analysis of chlorophyll
fluorescence transients for diagnosing the photosynthetic system of attached leaves. Plant Physiol. 84:
748-752.
[23] Omasa, K. (1996) Image diagnosis of photosynthesis in cultured tissues. Acta Hort. 319: 653-658.
[24] Genty, B. and Meyer, S. (1994) Quantitative mapping of leaf photosynthesis using chlorophyll
fluorescence imaging. Aust. J. Plant Physiol. 22: 277-284.
[25] Siebke, K. and Weis, E. (1995) Imaging of chlorophyll-a-fluorescence in leaves: Topography of
photosynthetic oscillations in leaves of Glechoma hederacea. Photosynth. Res. 45: 225-237.
[26] Meng, Q.; Siebke, K.; Lippert, P.; Baur, B.; Mukherjee, U. and Weis, E. (2001) Sink-source transition in
tabacco leaves visualized using chlorophyll fluorescence imaging. New Phytologist 151: 585-595.
[27] Oxborough, K. and Baker, N.R. (1997) An instrument capable of imaging chlorophyll a fluorescence
intact leaves at very low irradiance and at cellular and subcellular levels of organization. Plant Cell
Environ. 20: 1473-1483.
[28] Ibaraki, Y.; Iwabuchi, K. and Okada, M. (2004) Chlorophyll fluorescence analysis for rice leaves grown
under elevated CO 2 conditions. J. Agric. Meteorol. 60 (in press).
[29] Gitelson, A.A. (2004) Wide dynamic range vegetation index for remote quantification of biophysical
characteristics of vegetation. J. Plant Physiol. 161: 165-173.
[30] Chappelle, E.W.; Kim, M.S. and Mcmurtrey, J.E. (1992) Ratio analysis of reflectance spectra (RARS):
an algorithm for the remote estimation of the concentrations of chlorophyll a, chlorophyll b, and
carotenoids in soybean leaves. Remote Sens. Environ. 39: 239-247.
[31] Carter, G.A.; Rebbeck, J. and Percy, K.E. (1995) Leaf optical properties in Liriodendron tulipifera and
Pinus strobus as influenced by increased atmospheric ozone and carbon dioxide. Can. J. For. Res. 25:
407-412.
[32] Gamon, J.A.; Serrano, L. and Surfus, J.S. (1997) The photochemical reflectance index: an optical
indicator of photosynthetic radiation use efficiency across species, functional types, and nutrient levels.
Oecologia 112: 492-501.
[33] Yamamoto, H.Y. (1979) Biochemistry of violaxanthin cycle in higher plant. Pure Appl. Chem. 51: 639648.
[34] Stylinski, C.D.; Gamon, J.A. and Oechel, W.C. (2002) Seasonal patterns of reflectance indices,
carotenoid pigments and photosynthesis of evergreen chaparral species. Oecologia 131: 366-374.
[35] Carter, G.A.; Cibula, W.G. and Miller, R.L. (1996) Narrow-band reflectance imagery compared with
thermal imagery for early detection of plant stress. J. Plant. Physiol. 148: 515-522.
[36] Kozai, T.; Oki, H. and Fujiwara, K. (1990) Photosynthetic characteristics of Cymbidium plantlet in vitro.
Plant Cell Tissue Org. Cult. 22: 205-211.
[37] Fujiwara, K. and Kozai, T. (1995) Physical microenvironment and its effects. In: Aitken-Christie, J.;
Kozai, T. and Smith, M.A.L. (Eds.) Automation and Environmental Control in Plant Tissue Cultures.
Kluwer Academic Publishers, Dordrecht, The Netherlands; pp. 319-369.
[38] Fujiwara, K.; Kozai, T.; Nakajo, Y. and Watanabe, I. (1989) Effects of closures and vessels on light
intensities in plant tissue culture vessels. J. Agric. Meteorol. 45: 143-149 (in Japanese with English
abstract).
[39] Watanabe, S.; Nakano, Y. and Okano, K. (2001) Simple measurement of light-interception by individual
leaves in fruit vegetables by using an integrated solarimeter film. (Japanese text with English summary)
Environ. Control Biol. 39: 121-125.
29
www.taq.ir
MONITORING GENE EXPRESSION IN PLANT TISSUES
Using green fluorescent protein with automated image collection and analysis
JOHN J. FINER1, SUMMER L. BECK1,3, MARCO T.
BUENROSTRO-NAVA1,4, YU-TSEH CHI2,5 AND PETER P. LING2
1
Department of Horticulture and Crop Science, The Ohio State
University, 1680 Madison Ave., Wooster, OH 44691, USA – Fax: 330263-3887 – Email: [email protected]
2
Department of Food, Agricultural and Biological Engineering,
OARDC/The Ohio State University, 1680 Madison Ave., Wooster, OH
44691, USA
3
Current Address: DuPont Agriculture and Nutrition, Rt. 141 and Henry
Clay Road, Wilmington, DE 19880, USA
4
Current Address: IREGEP, Colegio de Postgraduados, Carretera
Mexico-Texcoco Km 35.5 Montecillo, Texcoco, Mexico, C.P. 56230
5
Current Address: 57 228 Lane Section 3 Yuanji Rd., Tianjhong Town,
Chang-Hua 520, Taiwan
1. Introduction
Automated systems are widely used across many discipline areas to perform tasks that
may be hazardous, time consuming, or impossible to perform by humans. In the plant
sciences, automated systems are being developed to execute difficult and tedious
activities and reduce the exposure of workers to agricultural chemicals [1].
In the area of plant developmental biology, automated systems have been developed
to gather information on how plants grow and develop under different environmental
conditions. Kacira and Ling [2] describe the use of a computer-controlled motorized
circular table and remote sensors to continuously monitor the health and growth of New
Guinea Impatiens plants growing under either low or high humidity conditions. An
infrared thermometer was used to collect data on the water stress index and a digital
camera was used to measure the top canopy area of the plants. Using this approach, it
was possible to detect the beginnings of a water deficit in the plants up to two days
before detection of visible wilting.
In the area of molecular biology, automated systems have tremendously improved
the capabilities of molecular biologists to perform complicated tasks with minimal
efforts. One of the first automated systems to receive widespread use in the area of
molecular biology is the thermocycler, which generates rapid temperature cycles,
31
© 2008 Springer.
www.taq.ir
www.taq.ir
J. Finer, S. Beck, M. Buenrostro-Nava, Y-T. Chi and P. Ling
enabling repeated synthesis of specific DNA fragments using a temperature insensitive
form of DNA polymerase. The Polymerase Chain Reaction (PCR) technique [3,4] has
revolutionized modern genetics by allowing efficient and accurate amplification of
DNA fragments from very small amounts of starting material. DNA sequencers are also
now fully automated and not only reduced the time and the labour required to obtain the
sequence of a certain DNA fragment, but have also provide insight into the genome of a
multitude of complex organisms. Genome sequencing is high throughput and both
sequence determination and alignment is automated.
One of the most recent applications of systems automation in the area of molecular
biology is the development of the microarray technology [5]. Microarrays are being
successfully used to assess the expression profile of thousands of genes from biological
samples [6-8]. For preparation of one type of microarray, thousands of small samples
are precisely placed on a microscope slide in an area generally of 3.5 by 5.5 mm. To
perform this fragile and laborious task, an automated system deposits multiple aliquots
of ~0.005 µl from thousands of different samples on a single slide. After fixation,
hybridization with fluorescent probe and washing, the slides are scanned with a laser
fluorescent scanner, which is equipped with a computer-controlled XY stage. To detect
the fluorescence, two photomultiplier tubes are used and the signal is split according to
the wavelength required to detect the fluorescence from each of the probes. The data is
processed and represented as an array, where each microscopic spot represents the
expression profile of the gene that was fixed at that particular point [5,9].
Although the use of microarray technology to profile expression of plant genes is
still relatively new, it has already become standard for high throughput analysis of gene
expression. Kazan et al. [6] used microarrays to screen 2375 Arabidopsis genes (based
on expressed sequence tags; ESTs), finding that 705 genes were up-regulated after the
plants were inoculated with a fungal pathogen or a signal compound. Comparisons of
the 705 genes with known sequences revealed that 106 of the genes had no previously
known function. Although microarray technology can be used to find new genes that are
up- or down-regulated under certain conditions, tissue extraction is required and precise
analysis of temporal expression can be difficult. Real-time analysis of gene expression
in living organisms is still useful, and visualization of transgene expression in living
tissue can provide additional information, that extracted tissue cannot.
2. DNA delivery
Although a number of different methods exist for introduction of DNA into plants cells
[10], particle bombardment [11] and Agrobacterium-mediated transformation [12] are
the two methods that have proven to be the most efficient and are most commonly used
by transformation laboratories for a large number of plant species.
2.1. PARTICLE BOMBARDMENT
For particle bombardment, DNAs are precipitated onto small (~1 µm) dense particles
(either tungsten or gold) and accelerated towards the target plant tissue, which is placed
under a partial vacuum to reduce drag on the particles. The particles are accelerated by a
blast of helium, released by either a fast-acting solenoid [13] or a rupture disc [14],
32
www.taq.ir
Monitoring gene expression in plant tissues
manufactured to rupture at specified helium pressures. Helium is used to propel the
particles as it is inert and possesses a high expansion coefficient. Once the particles
enter the target cells, the DNA is released from the particles, becomes associated with
the chromosomes and, if the proper conditions exist, the foreign DNA integrates into the
chromosomes of the target cell.
For particle bombardment, the DNA is physically delivered into the cells which
bypass any potential biological incompatibilities. But, the introduction of particles,
which range in size from 0.6 - 3 µm, can be damaging to the cells, which range in size
from 20 – 60 µm. To minimize damage, cells are often treated by physical or chemical
drying [15], which lowers the osmotic pressure in the cells and reduces the loss of
protoplasm through particle-generated holes in the cell wall.
Integrated DNA resulting from particle bombardment-mediated DNA transfer is
often high copy and fragmented [16,17] but this can be regulated by modifying the
introduced DNAs [18]. High copy transgenes can show variation or loss of expression
due to gene silencing [19].
2.2. AGROBACTERIUM
For Agrobacterium-mediated transformation, plant tissues are cultured in the presence
of Agrobacterium, which is a bacterium that has the unique ability to introduce part of
its DNA into plants [20]. Because Agrobacterium is a natural plant pathogen, some
biological incompatibilities exist when using certain plant species or stages of plant
growth. However, most of these biological incompatibilities have been removed or at
least lessened as more has been learned about the mechanism of DNA transfer [21].
With the addition of signal compounds [22] to the medium where Agrobacterium and
the plant tissues are co-cultivated, and enhancing exposure of cells to the invading
bacteria [23], the process of DNA transfer has become quite efficient for most plants.
Although antibiotics must be applied to eliminate the bacterium after DNA transfer,
this method of delivery has two distinct advantages over particle bombardment. First,
no instrumentation is required and the cost of performing DNA introductions is
minimal. Second, the DNA transfer process, which is mediated by the bacterium,
generally results in more consistent integration events. The transferred DNA (T-DNA)
is usually defined by specific borders and genes of interest can simply be engineered
between those borders. The resultant integrated DNA can be single copy or show
somewhat more complex integration patters [24].
3. Transient and stable transgene expression
Immediately following introduction, the fate of DNA can be inferred, based on early
events and eventual outcomes. Gene expression from the introduced DNAs can be
observed as early as 1.5 hours post-introduction [25] and is usually short-lived, lasting
1-3 weeks. This short-term expression is called, “transient expression” and probably
results from expression of DNA as an extrachromosomal unit. In addition, many of the
cells containing foreign DNA may not remain viable [26], due to the physical process of
DNA introduction or the response of the cells/tissue to invading bacteria. If the cells
remain viable following DNA introduction, the introduced DNA either degrades or
33
www.taq.ir
integrates into the DNA of the target cells. In plant cells, introduced DNAs are not
maintained as extrachromosomal elements. In most cases, once the DNA becomes
integrated, it becomes a stable transgenic event, resulting in “stable expression”. The
introduced T-DNA from Agrobacterium-mediated transformation is coated with protein
molecules and tagged with a protein signal peptide which assists with delivery to the
nucleus and integration into the chromosome [24]. Integration patterns in transgenic
plants obtained via particle bombardment-mediated DNA delivery suggests a high level
of recombination, resulting in a mixing rather than an insertion of the introduced DNAs
within the native plant DNA [27]. These recombination events most likely occur
directly following DNA introduction, during DNA integration into the chromosome.
Although the transition from transient to stable expression is very poorly
understood, it probably holds the keys to improving both transformation rates and
transgene expression. Studies of transient gene expression, directly following DNA
delivery along with a fine analysis of stable transgene expression are now possible
using the proper transgenic reporter genes and fine tracking of gene expression using
robotics and image analysis.
4. Green fluorescent protein
4.1. GFP AS A REPORTER GENE
Reporter genes have been developed and refined to “report” or visualize gene
expression in a variety of tissues and organisms. Early reporter genes coded for
enzymes, which required substrates which were converted into detectable or visible
forms following cleavage [28]. These early reporter genes worked well but substrates
were often costly and the assay itself could be toxic to the tissue, resulting in a single
time point determination of transgene activity. Today, the most commonly used reporter
gene is the Green Fluorescent Protein (GFP), which can be continually monitored over
time and does not require the use of a substrate as the protein product itself is
fluorescent. GFP has therefore become the most effective reporter gene for use in
transformation and for tracking gene expression.
The Green Fluorescent Protein is a naturally occurring protein found in jellyfish
(Aequorea victoria). The bioluminescence from this protein was first reported by
Ridgway and Ashley [29] and, since that first report, the use of green florescent protein
has expanded tremendously, impacting almost every field in the biological sciences;
especially plant sciences. This reporter gene has become increasingly useful for tracking
transgene expression in transformed plants.
Niedz et al. [30] first found that the wild-type gfp gene from the jellyfish could be
introduced into plant cells and visualized. The gfp gene has since been modified and
optimized to be the most effective reporter gene in plants. Wild-type GFP produces
green fluorescence expression at the wavelength of 507 nm (green) upon the excitation
at 395 nm (ultraviolet) or 475 nm (blue) [31]. In addition, sequence changes are usually
required when genes from organisms in one kingdom are transferred to organisms in
another kingdom. In plants, modifications to the gfp gene include the elimination of a
cryptic intron, alteration in codon usage, changes in the chromophore leading to
34
www.taq.ir
different excitation and emission spectra, and targeting to endoplasmic reticulum [32].
It has been developed as a reporter for gene expression, a marker of subcellular protein
localization, a tracer of cell lineage, and as a label to follow the development of
pathogens [33]. The GFP reporter allows detection of labelled protein within cells, and
monitoring of plant cells expressing GFP, directly within growing plant tissue [34].
Nagatani et al. [35] used digital imaging to monitor the heat shock response of
transgenic rice calli using GFP as a reporter gene. Images of transgenic calli were
acquired 0, 30, 60, and 120 minutes after heat treated for 10 min at 45°C. Analysis of
the images showed a 2-4-fold increase in the levels of GFP expression over time
compared to the control (no heat stress).
GFP has successfully been used as a reporter for evaluation of plant transformation
using both Agrobacterium [36] and particle bombardment [25]. GFP fluoresces under
blue light excitation, and it can be detected in as little as 1.5 hours following DNA
introduction [25]. Since GFP detection is non-destructive, expression can be followed
over extended periods of time using digital imaging [37].
Reporter genes provide an excellent way to not only examine gene expression but
also to evaluate expression over time in various tissues.
4.2. GFP IMAGE ANALYSIS
In the simplest terms, image analysis is the evaluation of an object using information
collected from an image. Image analysis can be totally manual or, at the highest level,
fully automated. For manual image analysis, the observer simply makes visual
judgments of the subject material and provides subjective qualitative ratings. At the next
level (interactive image analysis), images are collected and the operator assists with, but
does not complete the analysis of the images. The operator must separate the subject
from its background and demarcate or segment the region of interest in the field of view.
Input from the operation is therefore needed for every image, and objects or segments in
the images need to be outlined before the size/colour of the targets may be determined
using image analysis tools (i.e. blob analysis). Blob analysis groups pixels with the same
attributes (colour) into a region, which allows subsequent quantification of other factors
associated with the blob (width, length, area, etc). This interactive image analysis
process, although useful for some applications, is laborious and time consuming and is
not practical for high volume operations. At the highest level of image analysis,
automated quantitative analyses are performed. While automated image analysis is key
to high throughput monitoring of various subject materials, adaptive image analysis is
paramount to the success of analyzing images of varied quality. For adaptive image
analysis, the background, subject itself and regions of interest (ROI) within the subject
are separated. This can be challenging when images of varied colours and contrasts are
analyzed. To determine the percentage area of GFP expression in plant tissue, embryos
or plants, it is necessary to precisely identify the tissue, embryo or plant within the
image. In order to quantify GFP expression, it is also necessary to identify specific
regions in the target and determine if these regions are associated with GFP
fluorescence. Therefore, identification of the targets or blobs through “segmentation”
and parts of the target via “blob analysis” are needed for quantitative and qualitative
high throughput image analysis of GFP-expressing tissues. Currently, many evaluations
35
www.taq.ir
of gene expression and most assessments of tissue and plant quality still rely on human
vision, where results can often be highly variable and very subjective.
4.3. QUANTIFICATION OF THE GREEN FLUORESCENCE PROTEIN IN VIVO
With the widespread use of the gfp gene as a reporter gene, quantitative analyses of
GFP expression has been used to accurately gauge gene expression levels. Maximova et
al. [38] applied image analysis to quantify GFP expression in Agrobacterium-infected
leaf explants. Using the greyscale intensity of the area expressing GFP, intensity was
calculated from ten random areas of the subject. In this study, samples were visually
selected by the authors, which may have influenced the results. However, the potential
for utilizing image analysis for evaluating in situ GFP expression in plant tissues was
clearly demonstrated.
Hauser et al. [39] also used the average greyscale intensity of selected areas to
quantify the strength of GFP expression. The region of interest, which contained GFPexpressing Paramecium tetraurelia cells was selected randomly. Vanden Wymelenberg
et al. [33] analyzed the population of GFP-expressing Aureobasidium pullulans on leaf
surfaces using the average fluorescence per cell vs. cell number. Threshold values were
specified interactively to segment the region of interest from the background. Spear et
al. [40] used 256 scale levels to quantify GFP expression in fungal cells and obtained
intensity values using commercial image processing software. The region of interest
was segmented by simple ‘thresholding’, while the threshold value was selected by the
authors. The number of cells, individual cell areas, and total coverage area of the cells
were obtained by manual image analysis.
In order to achieve precise quantification of GFP expression, other variables, which
can change over time or between laboratories must be considered. Scholz et al. [41]
used an internal rhodamine B standard to correct the intensity fluctuations of the
exciting xenon arc lamp in the fluorescence spectrometer. Inoué et al. [42] quantified
GFP expression by calculating the average pixel intensity values of a circular region of
interest narrower than the samples. Since the strength of excitation light degraded with
time, GFP expression was corrected by subtracting the average background intensity
values of the region. The segmentation between the foreground and the background area
and the selection of either region of interest or the adjoining background was done
manually.
All of this research relied on manual input for image acquisition and image analysis.
An automated image collection and analysis system is desirable because of the time and
effort involved in collecting the and analyzing the images, which requires routine and
repeated manipulations and human involvement at numerous steps. For the monitoring
of a large number of targets, an automated system would insure higher efficiencies and
a greater consistency of high throughput data acquisition and analysis.
36
www.taq.ir
5. Development of a robotic GFP image acquisition system
5.1. OVERVIEW
Over the past few years, efforts in our laboratories have focused on assembly and
evaluation of an automated image acquisition for semi-continuous monitoring of GFP
expression in transiently- and stably-transformed plant tissues [43,44]. The automated
image acquisition system consists of a fluorescence dissecting microscope with a digital
camera and a custom-designed 2-dimentional robotics platform, all under computer
control (Figure 1). The total system was placed in laminar air flow hood and the hood
was housed in a temperature-controlled culture room for consistent temperature control.
The robotics platform was programmed to place the various samples, located in
different Petri dishes, under the objective of the microscope and the camera collected
the image before moving to the next target. The system presents unique problems due to
the aseptic nature of the tissue culture subject material and the “movement” of the tissue
due to tissue expansion and growth. Perhaps the greatest challenge was minimizing the
condensation on the lids of the sealed Petri dishes, which obscured the view of the
dishes’ contents and could make image analysis very inconsistent.
5.2. ROBOTICS PLATFORM
The robotics platform consisted of square piece of 5 mm thick Plexiglas measuring
about 40 cm x 40 cm. The platform was mounted on a 45 x 45 cm XY belt-driven
positioning table (Arrick Robotics Inc., Hurst, Texas) using 2 aluminium rails, which
were 5 cm tall and 40 cm long. The Plexiglas was sufficiently rigid to hold the samples
in place with no bending and the high transparency of this material minimized heat
buildup from absorbing the light used to illuminate the plant tissues. This was
problematic with earlier prototypes of the platform that were not transparent. Heat
accumulation within or on the platform causes the temperature of the dishes’ contents to
increase, leading to water condensation on the lid of the sealed Petri dishes.
Condensation reduces the quality of the images and makes the process of image
analysis difficult to impossible. To prevent heat accumulation on the bottom of the
platform, sixteen 6 cm diameter perforations were made in the Plexiglas, directly under
the eventual location of the Petri dishes. Small fans were initially mounted to the side of
the platform or in the 6 cm perforations but these were found to be unnecessary and
were not beneficial for elimination of condensation. But, these perforations were
retained as they did increase air flow. To secure the Petri dishes to fixed locations, a
mounting mechanism was incorporated into the platform design (Figure 1, inset).
The mounting mechanism was used to hold the dishes in place, suspend the dishes
over the platform surface, permit mounting of a black background material below the
dishes, and allow precise adjustment of the focal distances of different areas of a plate.
The mounting mechanism consisted, in part, of 3 plastic positioning screws which were
placed 120° apart from each other and 5 mm away from each 6 cm perforation (Figure
1, inset). One 100 x 25 mm Petri dish was placed on top of the tips of the three
positioning screws. As the tissue grew, the positioning screws were adjusted to maintain
focus of the subject materials. In addition, the positioning screws maintained the dishes
37
www.taq.ir
above the surface of the platform, permitting adequate air flow around the dish. A 7 cm
diameter piece of black card stock was placed on the head of the screws, suspended 1
cm below the platform surface. The black background provided a consistent background
for image analysis. To hold the Petri dishes in place, a 90° aluminium angle (2.5 cm
base and 2.5 cm high) was fastened to the platform and a plastic screw was horizontally
placed to press the plate against a polypropylene holder, which was cut to the same
shape as a Petri dish (Figure 1).
Figure 1. Automated image collection system showing the platform (P) mounted on the xy
belt-driven positioning table (XY). The weighted base (B) was needed to support the weight
of the microscope and camera, which were mounted on the long arm boom stand. The two
different light sources for this system were a halogen bulb (H), which provided white light
illumination, and a mercury bulb (M), which provided high energy blue light for GFP
detection. The mounting mechanism (inset) consists of 3 positioning screws and one
horizontal screw, which secured the Petri dish in place.
The platform was originally driven by two MD-2a dual stepper motors (Arrick Robotics
Inc.), each motor driving the movement in the X or Y direction. The table contained two
limit switches (one for each of the directions, X and Y), which were used to identify the
“home” position. This position was recognized by the computer when a limit switch
was activated by the platform. In order to place each sample under the microscope
objective, the platform was moved a specific number of steps from the home position in
the X and Y directions. The number of steps for each direction depended of the position
of the object on the platform.
Ideally, the robotics platform will place the subject in exactly the same location for
each image collection at each time point. Images, acquired at different times, should
present the same region for analysis. Time series images, having the same region of
analysis, guaranteed a precise dynamic quantification of GFP expression. Unfortunately,
this level of precision was not observed with this system. Positioning error was caused
by backlash of the drive belt and by the step losses from the motors. Unless the platform
was returned to the home position between each sample, the error accumulated and the
38
www.taq.ir
target tissue could actually move out of the field of view of the CCD camera if enough
points were taken prior to returning “home”. The error caused by backlash or losses of
motor steps occurred along both the X and Y axes of the positioning table.
Backlash error was reduced after replacement of the original motor system with
pulley reducers (PR23, Arrick Robotics, Hurst, Texas) and more powerful stepper
motors (MD-2b, Arrick Robotics, Hurst, Texas). This change reduced the motor step
size and increased the torque provided by the motors, improving the overall efficiency
of positioning. The smaller step size reduced the error caused by backlash and larger
torque reduced the possibility of step loss. This change did not eliminate backlash errors
completely, but it reduced the magnitude of the error. This improvement allowed
successive image collections of all of the samples within a single dish, and a return to
the home position was only required between dishes. This also reduced run times as it
was no longer necessary to return the platform to the home position between each
sample.
After the sample was positioned under the microscope objective, a 1 second delay
was used to minimize residual sample movement from the vibration caused by
repositioning of the platform. After saving the image, the platform was directed to the
next position within the same Petri dish or to the home position, if the next sample was
located in a different Petri dish.
5.3. HOOD MODIFICATIONS
The robotics system was placed in a custom-designed laminar air flow hood. A laminar
air flow hood was necessary as samples needed to be precisely placed in the dishes,
after the dishes were fixed in place using the mounting mechanism on the robotics table.
As a result, an aseptic environment was required. The basic hood design was an
isolation table style, where the hood working surface is physically separated from the
hood motors, thereby reducing or eliminating vibration from the hood motors. The table
of an isolation table style hood consists of a base table with a second platform,
suspended above the base table by rubber cushions. The second platform normally
consists of a laminate-covered surface, which was replaced by a similar-sized piece of
black epoxy lab counter top. Vibrations from the robotics system motors were reduced
or partially absorbed by the “vibration-free” work surface that the hood provides.
Because the image acquisition system was too tall to fit within standard hoods, the
working table was lowered to allow adequate clearance for the digital camera.
As this whole system was placed within a tissue culture room with lighted shelves
for growth of plant tissue cultures, light shielding was necessary. Extraneous light could
interfere with image analysis, especially when fluorescence was low. In addition, lights
in most laboratories are under photoperiod control and cycle on and off throughout an
image collection experiment. Light screens, consisting of wood frames covered in black
cloth and placed around the hood, were adequate but they were both bulky and
inefficient at light screening. The use of a curtain of black fabric, suspended from the
top of the hood opening was a simple and convenient solution to light leakage. The
curtain length was adjusted so that open space was present at the bottom of the curtain,
to allow free movement of the robotics platform. The air from the hood was able to
escape through this open space and it was found that the curtain also acted as a
temperature and air baffle, maintaining a more uniform temperature within the hood
39
www.taq.ir
space and reducing condensation on the lids of the Petri dishes even further.
Condensation of the lids of the dishes has been largely eliminated from additional
changes to Petri dish design (Finer, unpublished).
For long-term experiments requiring illumination, the standard fluorescent lights
mounted within the hood were replaced with Gro-lux™ fluorescent bulbs used in the
laboratory for growth of plant tissue cultures. These lights were placed under timer
control which allowed them to cycle on and off with a regular photoperiod, or the lights
could be automatically turned off during image collections.
5.4. MICROSCOPE AND CAMERA
A scientific charged-coupled device (CCD) SPOT-RT camera (Diagnostic Instruments
Inc., Sterling Heights, Michigan) was mounted on a Leica MZFLIII stereomicroscope
(Leica, Heerbrugg, Switzerland), which was mounted over the robotics platform using a
long arm boom stand. Due to the weight of the microscope and the camera, a heavy
weighted base was used with the long arm beam (Figure 1). The SPOT-RT camera was
selected for the automated system due to its high sensitivity to dim signals and the
flexibility to easily control basic functions such as gain, binning and exposure time. For
images collected using the unfiltered halogen bulb (see below), exposure times were
usually around one second. For collection of images showing GFP expression, exposure
times were as long as one minute. The proper exposure time for each of the channels
(red, green and blue) was predetermined for each type of image.
Digital images taken with the SPOT-RT camera could be represented in either 8 or
12 bits per pixel (bpp), which resulted in an intensity resolution of 256 or 4,096 discrete
grey levels, respectively, per pixel for each channel. Although colour images,
containing 12 bpp per colour channel, offer high resolution, they were seldom used
because their large size makes them difficult to store and analysis is very timeconsuming. To select the proper intensity resolution for the analysis of biological
samples, it is important to know the conditions in which the images need to be acquired.
Twelve bpp resolution images could be useful if it is difficult to distinguish objects
from their background. Images obtained with the SPOT-RT camera had a 32 bpp (8 bpp
per channel) resolution. The total memory size of each image was 5,760,054 bits for an
image size of 1600 x 1200.
5.5. LIGHT SOURCE AND MICROSCOPE OPTICS
To detect the expression of the GFP gene, the dissecting microscope was equipped with
a 100 W mercury bulb; with a “GFP-2” filter set, consisting of an excitation band pass
filter of 480/40 nm and a long pass barrier filter of 510 nm. The excitation filter allowed
the passage of the blue light produced by the mercury bulb, eliminating the light in the
UV, red and green spectra. The barrier filter blocked the blue light used to excite the
GFP and allowed observation of the green light emitted by the GFP. The barrier filter
allowed the passage of any visible light above the 510 nm spectrum, which was useful in
detecting fluorescence in other spectra. Green tissue, containing chlorophyll, fluoresced
red upon excitation with the high intensity blue light. It was also not unusual to observe
occasional yellow fluorescence in some tissues, from unknown compounds.
40
www.taq.ir
In addition to the mercury bulb light, the automated system also contained a 100 W
halogen lamp light source that was used to illuminate the objects under the microscope
with wide spectrum light. The light was transmitted from the light source to the object
through a glass fiber bundle to a 66 mm FOSTEC® (SCHOTT-FOSTEC LLC; Auburn
NY, USA) ringlight, which was attached to the objective of the microscope. This white
halogen light was useful when focusing the specimen and positioning the samples in the
centre of the field of view.
For experiments which did not require tracking of GFP expression, the halogen light
alone was used to illuminate the subject tissues, yielding sequential image collections
under white light. In this case, the filter set was not used and the halogen bulb was
automatically turned on for image collection only. In contract, for GFP image
collection, the mercury bulb remained on during the whole course of the experiment, as
the manufacturer recommended against continual re-starting of the bulb. With a bulb
life of 200-300 hours, long-term experiments were not possible. In addition, bulb
degeneration (30%) over the course of the experiment was expected, and controls were
necessary to detect and compensate for this loss of illumination intensity [44].
Experimental evaluation of custom-designed blue LED illuminators, which posses
much longer bulb lives, proved this light source inadequate for sufficient intensities of
illumination, even when 100 narrow angle LEDs were focused within a 1 cm field.
6. Automated image analysis
To measure plant growth and development, or to evaluate changes in GFP expression
accurately, the difference between two images, taken at different times, may be
determined by simply subtracting one from the other, providing that the two images
were taken under exactly the same conditions. Scaling, position, orientation and
illumination of targets in images taken at different times should be the same with this
automated image collection system.
6.1. IMAGE REGISTRATION
The automated image collection system described above provided close-to-optimal
conditions for automated image analysis. Magnification was constant although sample
positioning varied slightly. Positioning became more consistent with improved motors
on the robotics platform and the use of pulley reducers. Errors in positioning between
sequentially-collected images were corrected by an image registration operation along
the x and y axes. There were no orientation shifts observed in the target due to the
sample holder design.
Image registration is the process of aligning targets in an image series, using
mechanical or digital signal processing techniques. Re-alignment of images requires a
quantitative measurement of their similarity in order to determine the necessary
adjustments. Three similarity measures [45] were evaluated using images showing
transient gfp expression, collected using the automated image capture system. These
similarity measures are shown below.
41
www.taq.ir
Sum of the Absolute Value of Differences (SAVD):
W b H b
rm,n
¦¦ X
i, j
Yim, j n
i b j b
(1)
Correlation Function (CF):
W b H b
s m ,n
¦¦X
i, j
u Yi m, j n
(2)
i b j b
Correlation Coefficient (CC):
H
W
H
W
W
H
N
N
N
N
N
N
ª
º
2
2
2
2
2
«4 N 2 2
»
X
Y
(
X
)(
Y
)
u
¦
¦
¦
¦
¦
¦
i, j
i m, j n
i, j
i m, j n »
«
H
W
H
W
W
H
i
i
i
N j
N
N j
N
N j
N
2
2
2
2
2
2
¬«
¼»
U m, n
[4 N
2
W
N
2
H
N
2
¦ ¦
W
H
i
N j
N
2
2
X
2
i, j
W
N
2
(
H
N
2
¦ ¦X
2
i, j
) ][4 N
W
H
i
N j
N
2
2
2
W
N
2
H
N
2
¦ ¦
H
W
i
N j
N
2
2
2
i m, j n
Y
W
N
2
(
H
N
2
¦ ¦Y
i m, j n
)2 ]
H
W
i
N j
N
2
2
(3)
where X and Y are the two images to be registered. W and H are the width and height of
image X separately. m and n are the x and the y directional shift between image Y and
image X. Two images overlap completely when m and n are zero. r, s and ȡ are the
similarity matrices between two images. The value of element (m, n) in any of the
similarity matrix denotes the similarity of the two images when the shifts between the
two images in x and y direction are m and n. For the elements in matrix r, a lower value
means higher similarity. For the elements in matrices s and ȡ, a higher value means
higher similarity. The size of these similarity matrices depended on the range of m and
n. The range of m and n are determined by the maximum error which could occur in the
mechanical system. The range of i and j in the first 2 equations, which differ from m and
n (the x and y directional shift between two images), (region of calculation) are from b
to W – b and H – b, where ±b is the maximum and minimum shift in x and y direction,
respectively. The region of calculation guaranteed that every element in the similarity
matrix was calculated based on the same region of calculation. For example, when m =
0 and n = 0, the range of i and j could be from 0 to W and 0 to H in x and y direction,
which means the area of the region of calculation is WxH, because two images overlap
completely. When m = 50 and n = 50, the range of i and j could only be from 50 to W –
50 and H – 50 in x and y direction i.e. the area of the region of calculation is (W – 100)
x (H – 100) which is different from the previous case. Different region of calculation
may result in large error in finding the minimum or maximum value in those similarity
matrixes.
After evaluation of all three registration algorithms using artificially shifted images
showing transient GFP expression, it appeared that all 3 algorithms were capable of
precisely registering the images before and after the artificial shift regardless of the size
of the offsets.
42
www.taq.ir
The computational loads, required by the three methods, however, were significantly
different. Among the three algorithms evaluated, an average of 638 seconds was needed
for the CC method to register two 800 x 600 images. An average of 198 seconds and
255 seconds were required to register an image pair using the SAVD and CF measures
respectively. The computer used to evaluated the performance of the image registration
algorithms was a Pentium 4 2.0 GHz CPU personal computer with 384MB RDRAM
(Dimension 8200, Dell, Round Rock, Texas). SAVD was therefore found to be the most
efficient method to register images prior to GFP expression quantification.
6.2. QUANTIFICATION OF GFP
GFP expression can be quantified and presented in a number of different ways.
Analyses of transient expression have typically been presented as spot or foci counts
[11], which are usually based on counting GFP-expressing foci (which represent
individual GFP-expressing cells) by a human operator [25]. Foci counts are therefore
quite variable, depending on the individual counting the foci and their ability to discern
low intensity spots and minimize duplicate counting of foci in a crowded field.
However, counting foci is simple and does provide a good estimate of successful gene
introduction and an idea of the strength of the promoter used with the gfp gene. Using
automated image analysis, foci counts can be precisely and consistently quantified and
the intensity of GFP expression per focus or per sample can be easily determined.
To calculate the number of foci efficiently, blob analysis was applied to the binary
images following automated image registration. The advantage of blob analysis is its
computational efficiency. Blobs are areas of touching pixels that are in the same logical
pixel state i.e. grey scale level. It allows identification of connected regions of pixels.
The total numbers of blobs as well as the area of each blob in an image were obtained
using functions in a commercial image processing library (MIL, Matrox Inc., Quebec,
Canada). Fluorescence focus number per unit area was calculated using the equation
below.
Nn
Ns
Ai
(4)
where Nn is the foci number per unit area, Ns is the foci number calculated by blob
analysis and Ai is the area of the field of view (actual area analyzed) in mm2 after image
registration.
For quantification of GFP intensity, the average intensities in grey value of
foreground and background areas in the red and green spectra were calculated. For
determination of GFP expression per focus, the total grey value was divided by the
number of foci obtained by blob analysis.
7. Conclusions
Although the automated image collection and analysis system described in this chapter
is functional, problems exist in applying the technology to different target tissues.
43
www.taq.ir
For the robotics platform, samples must fit well within a Petri dish and rapidlygrowing tissues are exceedingly difficult to keep within the same focal plane.
Condensation on the lids of the Petri dishes has been largely controlled but the
temperature in the culture room, which contains the unit, does not fluctuate very widely
(± 0.5°C). This could be more of a problem in other laboratories, where environmental
control is less regimented. This system has taken 3 years to develop to the point of
functionality and it is not available commercially. The original dissecting microscope,
which was used to develop the system has been replaced by the manufacturer with a
modified design, which allows electronic focusing and automated exchange of filter
sets. Although this is very attractive, the complexity of the system would increase with
additional functionality. The automated image collection system does allow for the
collection of large amounts of images, which can be utilized for a number of different
purposes. The limiting factor for this work is in analyses and manipulation of the large
numbers of images that can be generated.
For image analysis of the collected images, semi-continual quantification of gene
expression and tissue growth has been possible. Quantification of promoter strength has
been shown and the potential of this system to characterize promoters and the factors
that induce gene expression should be evident. Growth of GFP-expressing organisms is
relatively easy to quantify [43] and the interaction of GFP-expressing organisms with
other organisms should assist in the study of some interactions. Additional applications
of this technology will undoubtedly arise, as it receives more widespread attention.
Individual images can be spliced together to yield time-lapse animations, which
allow compression of events and visualization of processes that have not been
previously observed. Time-lapse animations of tissue growth and expression of the gfp
gene provide additional information that will contribute to a greater understanding of
tissue growth and gene expression.
Acknowledgements
Salaries and research support were provided by State and Federal funds appropriated to
The Ohio State University/Ohio Agricultural Research and Development Centre.
Mention of trademark or proprietary products does not constitute a guarantee or
warranty of the product by OSU/OARDC, and also does not imply approval to the
exclusion of other products that may also be suitable.
References
[1] Kassler, M. (2001) Agricultural automation in the new millennium. Comput. Electron. Agric. 30: 237240.
[2] Kacira, M. and Ling, P.P. (2001) Design and development of an automated and non-contact sensing
system for continuous monitoring of plant health and growth. Trans. Am. Soc. Agric. Eng. 44: 989-996.
[3] Saiki, R.; Scharf, S.; Faloona, F.; Mullis, K.; Horn, G.; Erlich, H. and Arnheim, N. (1985) A novel method
for the parental diagnosis of sickle cell anemia. Am. J. Hum. Gene. 37: A172.
[4] Lee, M.S.; Chang, K.S.; Cabanillas, F.; Freireich, E.J.; Trujillo, J.M. and Stass, S.M. (1987) Detection of
minimal residual cells carrying the T-14 18 by DNA sequence amplification. Science 237: 175-178.
44
www.taq.ir
[5] Schena, M.; Shalon, D.; Heller, R.; Chai, A.; Brown, P.O. and Davis, R.W. (1996) Parallel human genome
analysis: Microarray-based expression monitoring of 1000 genes. Proc. Natl. Acad. Sci. - USA 93:
10614-10619.
[6] Kazan, K.; Schenk, P.M.; Wilson, I. and Manners, J.M. (2001) DNA microarrays: New tools in the
analysis of plant defense responses. Mol. Plant Path. 2: 177-185.
[7] Khan, J.; Wei, J.S.; Ringner, M.; Saal, L.H.; Ladanyi, M.; Westermann, F.; Berthold, F.; Schwab, M.;
Antonescu, C.R.; Peterson, C. and Meltzer, P.S. (2001) Classification and diagnostic prediction of
cancers using gene expression profiling and artificial neural networks. Nature Medicine 7: 673-679.
[8] Arcellana-Panlilio, M. and Robbins, S.M. (2002) Cutting-edge technology I. Global gene expression
profiling using DNA microarrays. Am. J. Physiol. 282: G397-G402.
[9] Eisen, M.B. and Brown, P.O. (1999) cDNA Preparation and Characterization. In: Weissman, S. (Ed.)
DNA arrays for analysis of gene expression. Academic Press, New York; pp.179-205.
[10] Finer, J.J.; Finer, K.R. and Santarem, E.R. (1996) Plant cell transformation, physical methods for. In:
Meyers, R.A. (Ed.) Encyclopedia of Molecular Biology and Molecular Medicine. VCH Publishers, The
Netherlands; pp. 458-465.
[11] Klein, T.M.; Wolf, E.D.; Wu, R. and Sanford, J.C. (1987) High-velocity microprojectiles for delivering
nucleic acids into living cells. Nature 327: 70-73.
[12] Horsch, R.B.; Fry, J.E.; Hoffman, N.L.; Eicholtz, D.; Rogers, S.G. and Fraley, R.T. (1985) A simple and
general method for transferring genes into plants. Science 227: 1229-1231.
[13] Finer, J.J.; Vain, P.; Jones, M.W. and McMullen, M.D.(1992) Development of the Particle Inflow Gun
for DNA delivery to plant cells. Plant Cell Rep. 11: 232-238.
[14] Sanford, J.C.; DeVit, M.J.; Russell, J.A.; Smith, F.D.; Harpending, P.R.; Roy, M.K. and Johnston, S.A.
(1991) An improved, helium-driven biolistic device. Technique 3: 3-16.
[15] Vain, P.; McMullen, M.D. and Finer, J.J. (1993) Osmotic treatment enhances particle bombardmentmediated transient and stable transformation of maize. Plant Cell Rep. 12: 84-88.
[16] Hadi, M.Z.; McMullen, M.D. and Finer, J.J. (1996) Transformation of 12 different plasmids into soybean
via particle bombardment. Plant Cell Rep. 15: 500 -505.
[17] Kohli, A.; Griffiths, S.; Palacios, N.; Twyman, R.M.; Vain, P.; Laurie, D. and Christou, P. (1999)
Molecular characterization of transforming plasmid rearrangements in transgenic rice reveals a
recombination hotspot in the CaMV 35S promoter and confirms the predominance of microhomology
mediated recombination. Plant J. 17: 591-601.
[18] Fu, X.; Duc, L.T.; Fontana, S.; Bong, B.B.; Tinjuangjun, P,; Sudhakar, D.; Twyman, R.M.; Christou, P.
and Kohli, A. (2000) Linear transgene constructs lacking vector backbone sequences generate low-copynumber transgenic plants with simple integration patterns. Trans. Res. 9: 11-19.
[19] Napoli, C.; Lemieux, C. and Jorgensen, R. (1990) Introduction of a chimeric chalcone synthase gene into
petunia results in reversible co-supression of homologous genes in trans. Plant Cell 2: 279-289.
[20] Chilton, M.D.; Drummond, M.J.; Merlo, D.J.; Sciaky, D.; Montoya, A.L.; Gordon, M.P. and Nester,
E.W. (1977) Stable incorporation of plasmid DNA into higher plant cells: the molecular basis of crown
gall tumorigenesis. Cell 11: 263-271.
[21] Hansen, G.; Das, A. and Chilton, M.D. (1994) Constitutive expression of the virulence genes improves
the efficiency of plant transformation by Agrobacterium. Proc. Natl. Acad. Sci.- USA 91: 7603-7607.
[22] Stachel, S.E.; Messens, E.; Van Montagu, M. and Zambryski, P. (1985) Identification of the signal
molecules produced by wounded plant cells which activate the T-DNA transfer process in Agrobacterium
tumefaciens. Nature 318: 624-629.
[23] Trick, H.N. and Finer, J.J. (1997) SAAT: Sonication Assisted Agrobacterium-mediated Transformation.
Transgenic Res. 6: 329-336.
[24] Zupan, J.; Muth, T.R.; Draper, O. and Zambryski, P. (2000) The transfer of DNA from Agrobacterium
tumefaciens into plants: a feast of fundamental insights. Plant J. 23: 11-28.
[25] Ponappa, T.; Brzozowski, A.E. and Finer, J.J. (1999) Transient expression and stable transformation of
soybean using the jellyfish green fluorescent protein (GFP). Plant Cell Rep. 19: 6-12.
[26] Hunold, R.; Bronner, R. and Hahne, G. (1994) Early events in microprojectile bombardment: cell
viability and particle location. Plant J. 5: 593-604.
[27] Svitashev, S.K.; Pawlowski, W.P.; Makarevitch, I.; Plank, D.W. and Somers, D.A. (2002) Complex
transgene locus structures implicate multiple mechanisms for plant transgene rearrangement. Plant J. 32:
433–445.
[28] Jefferson, R.A. (1987) Assaying chimeric genes in plants: the GUS gene fusion system. Plant Mol. Biol.
Rep. 5: 387-405.
45
www.taq.ir
[29] Ridgway, E.B. and Ashley, C.C. (1967) Calcium transients in single muscle fibers. Biochem. Biophys.
Res. Commun. 29: 229–230.
[30] Niedz, R.P.; Sussman, M.R. and Satterlee, J.S. (1995) Green Fluorescent protein: an in vivo reporter of
plant gene expression. Plant Cell Rep. 14: 403-406.
[31] Stewart, C.N. (2001) The utility of green fluorescent protein in transgenic plants. Plant Cell Rep. 20:
376-382.
[32] Haseloff, J. and Amos, B. (1995) GFP in plants. Trends in Genetics 8: 328-329.
[33] Vanden Wymelenberg, A.J.; Cullen, D.; Spear, R.N.; Schoenike, B. and Andrews, J.H. (1997)
Expression of green fluorescent protein in Aureobasidium pullulans and quantification of the fungus on
leaf surfaces. BioTechniques 23: 686-690.
[34] Haseloff, J. and Siemering, K.R. (1998) The uses of green fluorescent protein in plants. In: Chalfie, M.
(Ed.) Green Fluorescent Protein: Properties, Application, and Protocols. Wiley-Liss, Inc., New York; pp.
191-219.
[35] Nagatani, N.; Takuni, S.; Tomiyama, M.; Shimada, T. and Tamiya, E. (1997) Semi-real time imaging of
the expression of a maize polyubiquitin promoter-GFP gene in transgenic rice. Plant Sci.124: 49-56.
[36] Grebenok, R.J.; Lambert, G.M. and Galbraith, D.W. (1997) Characterization of the targeted nuclear
accumulation of GFP within the cells of transgenic plants. Plant J. 12: 685-696.
[37] Piston, D.W.; Patterson, G.H. and Knobel, S.M. (1999) Quantitative imaging of the green fluorescent
protein (GFP). In: Methods in Cell Biology, Nashville, Tennessee; pp. 31-48.
[38] Maximova, S.N.; Dandekar, A.M.; and Guiltinan, M.J. (1998) Investigation of Agrobacterium-mediated
transformation of apple using green fluorescent protein: high transient expression and low stable
transformation suggest that factors other than T-DNA transfer are rate-limiting. Plant Molec. Biol. 37:
549 – 559.
[39] Hauser, K.; Haynes, W.J.; Kung, C.; Plattner, H. and Kissmehl, R. (2000) Expression of the green
fluorescent protein in Paramecium tetraurelia. Eur. J. Cell Biol. 79: 144-149.
[40] Spear, R.N.; Cullen, D. and Andrews, J.H. (1999) Fluorescent labels, confocal microscopy, and
quantitative image analysis in study of fungal biology. In: Methods in Enzymology, Vol. 307: 607-623.
[41] Scholz, O.; Thiel, A.; Hillen, W. and Niederweis, M. (2000) Quantitative analysis of gene expression
with an improved green fluorescent protein. Eur. J. Biochem. 267: 1565-1570.
[42] Inoué, S.; Shimomura, O.; Goda, M.; Shribak, M. and Tran, P.T. (2002) Fluorescence polarization of
green fluorescence protein. Proc. Natl. Acad. Sci. -USA 99: 4272-4277.
[43] Buenrostro-Nava, M.T.; Ling, P.P. and Finer, J.J. (2003) Development of an automated image collection
system for generating time-lapse animations of plant tissue growth and green fluorescent protein gene
expression. In: Vasil, I.K. (Ed.) Plant Biotechnology 2002 and Beyond. Kluwer Academic Publishers,
The Netherlands; pp. 293-295.
[44] Buenrostro-Nava, M.T.; Ling, P.P. and Finer, J.J. (2005) Development of an automated image
acquisition system for monitoring gene expression and tissue growth. Trans. Am. Soc. Agric. Eng. (in
press).
[45] Svedlow, M.; McGillem, C.D. and Anuta, P.E. (1978) Image registration: similarity measure and preprocessing method comparisons. Aerospace and Electronic Systems, IEEE Trans. AES 14: 141-149.
46
www.taq.ir
APPLICATIONS AND POTENTIALS OF ARTIFICIAL NEURAL NETWORKS
IN PLANT TISSUE CULTURE
V.S.S. PRASAD AND S. DUTTA GUPTA
Department of Agricultural and Food Engineering, Indian Institute of
Technology, Kharagpur 721 302, India – Fax: 91-3222-255303 –
1. Introduction
In a broad sense, intelligence is something, which deals with the ability to grasp,
analyze a task and then reach for a logical conclusion upon which an action can be
initiated. Over the years, many researchers have been attempting to create a nonbiological entity that can match human level performance. Such attempts have
manifested in the emergence of a cognitive approach termed as artificial intelligence
(AI). There are many ways in which artificial intelligence can be manoeuvred to
execute its function. Computers can be programmed to provide a platform for a
coherent approach for executing a particular task. Complex mathematical functions can
be deciphered and logical theorems can be deduced by the use of symbolic artificial
intelligence. But symbolic artificial intelligence neither could decrypt a digitized image
nor could deduce a signal from imperfect data, and has difficulty in adapting things to a
change in a specified process. Many problems do exist which cannot be elucidated by
simple stepwise algorithm or a precise formulae, particularly when the data is too
complex or noisy. Such problems require a sort of connectionism or in other words a
network approach. It is possible to interconnect many mathematical functions, all of
which perform a dedicated task of processing the data into a desired form of meaningful
output. The data can be forwarded through valued connection routes. The conduction
strength of the routes, which regulates the movement of data processing can act as a sort
of memory and can be very useful in adapting to process changes. Function wise, such
network approach is exactly the reverse of symbolic AI. The strength of neural network
analysis lies in its ability to generalize distorted and partially occluded patterns and
potential for parallel processing. However, they encounter difficulty in formal reasoning
and rule following. The results of applying such network technology have been found to
be astounding and phenomenal with a relatively modest effort.
Biological processes are incomprehensible in terms of their behaviour with respect
to time. It is a well-recognized fact that the genetic and environmental factors are the
key effectors, which contribute to their functioning. These two factors have a very high
degree of variability in and among themselves ultimately manifesting in a wide spectrum
of biological developments that are non-deterministic and non-linear in nature. Such
47
© 2008 Springer.
www.taq.ir
V.S.S. Prasad and S. Dutta Gupta
developmental patterns are also characteristic to plant cells and tissues, which are
cultured aseptically in controlled but stressful in vitro environment. In vitro plant
culture practice is generally intended to manipulate the tissue growth and behaviour in a
predefined manner either to obtain mass propagated elite plantlets within a short
timeframe or to derive useful metabolites on a large scale apart from its use in transgene
research. Appropriate modelling which can predict as well as simulate in vitro growth
kinetics, thermodynamic limitations of culture conditions and energy to mass and vice
versa conversions in a realistic manner are therefore considered very much essential.
Conventional analytical techniques for these purposes based on mathematical models
are questionable, since these methods do not conform to the non-idealities of in vitro
culture phenomenon.
Neural network technology is an efficient alternative for reliable and objective
evaluations of the biological processes. Neural network technology deals with
approximation of different complex mathematical functions to process and interpret
various sets of erratic data. This technology mimics the structure of the human neuron
network as it incorporates information processing and decision making capabilities.
With their high learning capability, they are able to identify and model complex nonlinear relationships between the input and output of a bioprocess [1,2,3]. While, neural
networks have shown remarkable progress in the area of on-line control of bioprocesses,
their applications to complex plant tissue culture systems are comparatively recent and
restricted only to a few instances.
The present chapter primarily aims to introduce the fundamental concepts of
artificial neural network technology to those who own more of an authentic command in
life sciences than in mathematics and allied fields. This review is intended to explain
the relevance of network based evaluations in plant cell and tissue culture as compared
to conventional syntactic approaches, discuss basic methodology of network modelling,
describe the various applications of artificial neural networks in in vitro plant culture
systems and provide an insight into the future perspective and potentials of network
technology.
2. Artificial neural networks
2.1. STRUCTURE OF ANN
The fundamental structure of ANN is similar to that of a biological nervous system. The
network architecture is a connected assembly of individual processing elements called as
nodes. These nodes are arranged in the form of layers. The most common structure is a
three-layered network as depicted in Figure 1. It comprises of an input layer, a hidden or
interactive layer and an output layer. A three-layered network is shown as an example
because it can address all the problems that a more complex network is capable of
though not as efficient. The connections between nodes and the number of nodes per
layer are defined by the approach, which is adopted to solve or interpret a given
problem. The flow of the information through a network is governed by the direction of
inter-nodal connections. In feed forward neural network, unidirectional connections
exist between the neurons belonging to either same or different layers allowing the
48
www.taq.ir
Applications and potentials of artificial neural networks in plant tissue culture
processed data proceeds only in forward direction, whereas in Recurrent neural network
(Feed-back network) connections exist in both forward and backward direction between
a pair of neurons and even in some cases from a neuron to itself.
Figure 1. Three layered feed-forward network.
2.2. WORKING PRINCIPLE AND PROPERTIES OF ANN
2.2.1. Computational property of a node
The functioning of individual node in ANN is analogous to the biological neuron. Each
node receives one or multiple inputs from surrounding node(s) and computes an output
that is transmitted to the next node. While computing the output, the input information
is weighed either positively or negatively. Assigning some threshold value to the
concerned neuron simulates the output action. At the level of each node, the input
values are multiplied with the weight associated with the input to give a result. The
result is then adjusted by an offset variable `ș’ according the type of network in use.
The output is then determined using the adjusted summation as the argument in a
function `f’ which is pre-defined by the algorithm (Figure 2). Function ‘f’ can also be
termed as either transfer function or activation function. This function can take sigmoid,
linear, hyperbolic tangent or radial basis form. The selection of the activation function
depends on the purpose of the network.
49
www.taq.ir
Figure 2. Basic mechanism of nodal computation n = No. of inputs; x = input variable;
w = weight of ith input; ș = internal threshold value and f = activation function.
The most common neuronal nonlinear activation function used in biological systems is
sigmoid in nature (Figure 3).
Figure 3. Sigmoid activation function.
Figure 4. Steps of neural computation.
50
www.taq.ir
The ability of the network to memorize and process the information lies in the weights
assigned to the inter-node connections, which determine their conductivity through a
network. These weights are incurred during the process of training the network. The
inter-nodal connections with their corresponding weights basically represent the
adaptability of the network to the problem domain. When input variables are fed to the
neural network, the corresponding computed outputs are compared to the desired output
(as in the case of supervised training mechanism). The error thus generated is
propagated back to the network for some parametric adjustments (also called as learning
rule) until the network attains a good generalization of the problem domain (Figure 4).
2.2.2. Training mechanisms of ANN
One of the major properties of the neural networks is to learn and adapt to input
information to produce convincing results. Many different training mechanisms have
been incorporated in neural networks. Training mechanism also influences the speed
with which the network converges and affects the accuracy of models, which classify
unknown cases. The ANN learns either in supervised or unsupervised fashion. In
supervised method, the external `conductor’ provides the desired output values that are
then matched to the system output values for the purpose of correcting the network
functioning. In unsupervised method, the system develops its own representation of the
input stimuli. For example in pattern classification self-organizing network, the system
autonomously recognizes the statistically salient features of the input patterns and
categorizes them. Unlike the supervised learning paradigm, there are no pre-determined
sets of categories into which the patterns are to be classified.
2.3. TYPES OF ARTIFICIAL NEURAL NETWORKS
Neural networks can be differentiated either based on the purpose for which they are
devised or on their basic topology along with the associated training method. Since our
interest is to describe the applicability of ANN to plant tissue culture systems, we
restrict only to the types of models with respect to their applications.
2.3.1. Classification and clustering models
ANN can be used for pattern recognition, nonlinear regression and classification
purpose in plant tissue culture studies. For automation in commercial mass propagation
of plants, decision-making networks play a major role, which come under this category.
Classification models find most common application in tissue culture. Multilayer
Perceptron (MLP) [4], Backpropagation neural networks (BPNN) and ADALINE
networks comprise the categorization networks with supervised learning. Unsupervised
architectures rely mostly on the data for clustering the input patterns. Under this
category Kohonen network, Competitive and Hebbian learning, Adaptive resonance
theory (ART) can be placed. BPNN are well suited for pattern matching and trend
analysis. It is just like feed-forward neural network. In order to adjust the connection
weights from input to hidden nodes, the errors of the units in the hidden layers are
determined by back propagating the errors of the units of the output layers in a
supervised manner. This is also called as back-propagation learning rule. Such neural
networks are called back propagation neural networks.
51
www.taq.ir
2.3.2. Association models
These are the models, which accept binary valued inputs. These neural networks
associate an object by just `seeing’ a part of that object. Continuous variables in such
cases can be converted to binary form to be used as input. These models endorse
threshold approach. In this case, the neurons are never connected to themselves.
Hopfield networks, Binary associative memory (BAM), Adaptive binary associative
memory (ABAM) and Hamming networks are examples of this type.
2.3.3. Optimization models
In plant tissue culture studies, there is a need to optimize the process taking into account
the factors influencing them. Optimization models find a best solution when trained
with a set of constraints. The weights of these constraints are stored in the connections
so that when independent variables are fed the network predicts the combination of
variables that would yield optimum solution.
2.3.4. Radial basis function networks (RBFN)
These networks endorse a combination of supervised and unsupervised learning
methods. They are mainly used for modelling a biological process, classification and
reduction in the dimensionality of the process. In this type of architectures, the hidden
layer is trained by unsupervised learning methodology like for example K-means
algorithm, whereas the output nodes are modelled based on supervised learning like for
example least mean square algorithm. In RBFN, centres are located among the input
and output pairs. A good generalization is represented by minimum values of sum of
squares of the distance between the centres to training data sets. In other words, the
activation function of each node uses a distance measure as an argument. It is very
much applicable to function approximation problems. RBFN are easy to work with and
are very fast `learners’ and show good generalizations and classifications. They are
good for image recognition. It is just like BPNN with similar kind of information flow.
2.4. BASIC STRATEGY FOR NETWORK MODELLING
The model of the neural network to be used depends largely on our purpose. The type of
the network affects the required form and quality of output.
2.4.1. Database
The neural computation is largely dependent on the availability of the data sets. Neural
network modelling is appropriate if the database is complete (data representing all the
aspects of the subject). Network approach can also be adopted in case of incomplete
database provided an expert opinion is available (as in the case of supervised learning).
In network modelling the variability in the data sets is more important than its
availability in large quantity. While obtaining the data the meaningful parameters must
be chosen which hold significant relevance to the purpose of modelling. Some ANN
accepts binary data while others accept continuous variables as inputs. In plant tissue
culture studies, information can be obtained form the following data types:
52
www.taq.ir
x
x
x
x
Binary data (organogenic / non-organogenic; viable / non-viable; regenerable /
recalcitrant)
Continuous (growth rate)
Categorical (growth regulator treatment categories; poor, moderate and good
response)
Fuzzy (the degree of hyperhydricity)
While selecting an approach, relevant data must be scored in a suitable format with
regard to the type of application one intends to develop. The information can be
encoded into one of the data types before feeding depending on the type of output one
can expect. The sensitivity of the output pattern to a particular input pattern varies not
only with the value of that input, but also with the values of the other accompanying
inputs. Therefore, the independent input variables should be scaled to the same range or
same level of variance before they are fed to the network. Categorical variables must be
ordered either in ascending or descending form. If the data is incomplete, to ensure the
integrity of the information, one can enter both minimum and maximum values or enter
average values taking into account its specific impact on the output quality. For online
process monitoring and decision control, data can be obtained in the form of time series.
In such systems, in order to avoid data overload and to accomplish real-time
interpretation, proper sampling rate must be determined to keep the data points to
minimum without loosing crucial information. Data can also be decoded from digitized
images using appropriate image software to render image information amenable for
neural computation.
For optimal performance of the ANN, the size of the training data set is very
important since ANN derives its information from the input data sets. The training data
sets should represent full range of conditions, so that the network defines a subjected
system in a comprehensive manner. The training sets should be always greater than the
number of weights in ANN. A preferred size of the training set is 3 to 10 times that of
the number of weights. If we train the network with small number of learning data set,
initially the error in the output will be very high. But as and when the learning iterations
are continued, the error in the learning set tends to decrease. The process of training is
stopped when the output error does not decrease anymore but contrarily shows as
increasing trend. When the network output goes perfectly through the learning samples,
the error with the learning set is least. However, when test data set is fed to such trained
network, the error would be very high. The average learning and test error rate is a
function of the learning data set size. The learning error increases with an increasing
learning set size, and the test error decreases with increasing learning set size. A reliable
network performance is evaluated based on smaller test error than on the learning error.
With increasing number of learning sets the error rates of learning and test sets converge
to the same value at some point and at that point the learning procedure attains a good
approximation.
2.4.2. Selection of network structure
Generally input and output nodes are fixed as per the necessity and one hidden layer
would be sufficient for estimating any non-linear biological function. More than one
interactive hidden layer can be incorporated when different layers comprising hidden
53
www.taq.ir
nodes have different task to perform as in the case of Hypernet algorithm. Apart from
the number of layers the connectivity between the nodes affects the functioning of the
network. The size of the training set and the interpretation of the output are dependent
on the inter-nodal connectivity.
2.4.2.1. Number of input nodes. The number of nodes in the input layer must
correspond to the number of variables that are taken into account. An expert can fix the
number of nodes in input layer based on the relevancy of the corresponding variable.
ANOVA can be performed to select statistically significant variables and nodes can be
assigned to them. Threshold based selection of input nodes can also be done. That is
when the weights during learning drop below a threshold level or nearly equals to zero,
the nodes associated with them may be pruned accordingly. Combination of input
variables that are highly correlated can also lead to justified inclusion of the input
nodes.
2.4.2.2. Number of hidden units. Error criteria based upon the number of learning
iterations is then taken into account to determine how many processing elements should
be there in the hidden layer. When large number of hidden nodes is considered, the
network fits exactly with the learning data sets. However, the function the network
represents will be far wayward because of the extensive connectivity with both input
and output layers. Particularly in case of learning data sets derived from biological
experimentations, which contain a certain amount of noise, the network will tend to fit
the noise of the learning samples instead of making a smooth and meaningful
approximation. It has been shown that a large number of hidden nodes lead to a small
error with the training set but not necessarily lead to a small error in the test set. Adding
hidden units will always lead to a reduction of the error during learning. However, error
on test sets initially gets reduced as hidden nodes are added, but then gradually increase
if more than optimum hidden nodes are incorporated per layer (Figure 5). This effect is
termed as the peaking effect. The architecture that gives smallest error is normally
selected as the best choice.
2.4.2.3. Learning algorithm. Once the topology of the network is selected, the choice of
the learning algorithm will be automatically gets defined. Learning algorithm is greatly
dependent upon the type of input nodes (binary, continuous or fuzzy) and also the internodal connectivity. Learning algorithm also influences the network convergence ability
and its stability. Some learning algorithms may be unstable in some conditions.
Therefore, certain limiting conditions must be specified. The algorithm must be
appropriate for the type of input data and should be able to produce desired form of
output. Algorithms that demand higher number of iterations pose problems in
propagating the error.
54
www.taq.ir
Figure 5. Effect of number of hidden nodes on output precision.
The following aspects need to be considered, while training an algorithm:
x Time required for training
x Number of iterations required
x Convergence of the algorithm
x Stability of the solutions when additional vectors are added to training set
x Stability of solution when the order of training vectors is altered.
The most common learning algorithm is backpropagtion method. Here, the error that is
generated due to discrepancies between the system output and the expected outcome is
propagated back to facilitate readjustments of the weights assigned to the connections
till the network achieves a good generalization.
2.4.3. Training and validation of the network
If there is `N’ number of experimental data representing different conditions it has to be
determined whether the data should be presented to the network one set at a time
(sequential) or all the data in the matrix form and then processed in parallel (parallel).
Sequential training is considered best because when the network converges using a
particular data set, the weights are saved and are used as initial weights for the next data
set and so on which is not possible in parallel processing. Fundamental aspect of training
ANN is the stop criterion, which implies the point at which the training is terminated.
The error in the training set tends to decrease with training iterations when the ANN has
enough degrees of freedom to represent the input/output map. After such training of the
network, the validity of the network is tested. Finally a cross validation of error is
obtained for different topologies comprising of different number of hidden nodes to
minimize error in network response. Smaller number of nodes will cause the ANN to be
insufficiently flexible to represent the experimental signal and too many nodes will
allow an excess of degrees of freedom which will cause premature over-fitting and
consequently, cross-validation will terminate the learning process for a higher error. An
55
www.taq.ir
alterative way to control it is to reduce the size of the network. Either one can set a small
topology with fewer hidden nodes and add new nodes or can begin with a large network
and remove the nodes to get minimum error with test set [5]. To avoid over-training or
over-fitting (a condition where the ANN strongly remembers only the training patters),
the performance obtained with the validation set must be checked once in every 50
passes of the training set. The validation step should comprise at least 10% of the
training steps and the data set of the validation must be distinct from the training set.
3. Applications of ANN in plant tissue culture systems
Plant tissue culture is an excellent technique for commercial mass propagation of elite
plant species in a relatively short period of time overcoming the limitations poised by
agro-climatic, seasonal and biotic effects on conventional plant production
methodologies. Large-scale cultivation of plant cells in bioreactor has also been found
effective for production of high value natural compounds. However, developmental
pattern of somatic embryos, characteristics of regenerated plants and behaviour of in
vitro cell cultures makes the conventional modelling technique ineffective for on-line
monitoring. ANN can be leveraged to plant tissue cultural practices for pattern
recognition of somatic embryos, photosynthetic and photometric evaluation of
regenerated plants and on-line evaluation of biomass and control of secondary
metabolite production. ANN based modelling approach has been found to be more
flexible, effective and versatile in dealing with non-linear relationships prevalent in cell
culture practices. Also the approach has distinct advantages, as it does not require any
prior knowledge regarding the structure or interrelationships between input and output
signals. The various applications of ANN in plant tissue culture systems are
summarized in Table 1. These studies provide a comprehensive insight into the
expediency of processing networks in interpreting the database derived from in vitro
plant culture investigations.
3.1. IN VITRO GROWTH SIMULATION OF ALFALFA
This case study deals with the simulation of in vitro shoot growth of alfalfa for
transplant production [6]. Combined effects of CO2 inside the culture vessel and sucrose
content of the media on in vitro shoot growth were studied. A growth model using
Kalman filter neural network was developed for this purpose. The experimental data of
growth parameters such as dry weight, leaf number and root initiation stage were
correlated well with the simulated values calculated by the trained neural network.
The study demonstrates the efficacy of Kalman filter training of the neural network
in simulation of in vitro plant growth. This pioneering work also laid a foundation
towards an entirely divergent method of understanding the in vitro plant growth, which
usually tends to behave in a non-deterministic way.
56
www.taq.ir
Table 1. Applications of artificial neural network in plant tissue culture studies.
Application
Network model
Associative
technique
employed
Database source
References
Growth simulation of
alfalfa shoots as
effected by CO2 and
sucrose levels
Neural network
with Kalman filter
training method
Growth
modelling
Dry weight, leaf
number and root
initiation stage
[6]
Distinguishing different
embryo types from
non-embryos and
predicting embryo
derived plantlet
formation
Feed-forward
Image analysis
Area, length to width
ratio, circularity and
distance dispersion
of plant cell cultures
[7]
Biomass estimation of
cell cultures
Standard feedforward neural
network with
gradient descent
method of
optimization and
sigmoid function
as neuron
activation
Quick basic
programming of
algorithm
Sucrose, glucose and
fructose level of
medium
[8]
Simulation of
temperature distribution
in culture vessel
Three layered
neural network
trained with
Kalman filter
Finite element
formulation
programmed in
Visual Basic3.0
Spatial temperature
distribution of
culture vessel
[9]
Identification and
estimation of shoot
length
Fuzzy neural
network with back
propagation
algorithm and
sigmoid function
of neurons
Image analysis
and multiple
regression
modelling;
algorithms
programmed in
VC++ language
Pixel brightness
values in red blue
and green colour
regimes
[10]
Classification of
somatic embryos
Feed-forward
neural network
Image analysis
and discrete and
fast Fourier
transformation
Radius, length,
width, roundness,
area and perimeter of
the somatic embryo
images
[11]
Clustering of
regenerated plant-lets
into groups
Adaptive
resonance theory 2
Image analysis;
`C’ language
based
programming
Mean brightness
values, Maximum
pixel count and grey
level of maximum
pixel count in RBG
regions
[12]
57
www.taq.ir
3.2. CLASSIFICATION OF PLANT SOMATIC EMBRYOS
The germination and conversion frequency of somatic embryos depend on the normalcy
and the developmental stage of the normal embryo. Hands-on selection of such
embryos, though accurate, is very laborious, time consuming (since the embryos take
approximately eight weeks to mature) and cost intensive. Therefore, an efficient
automated system is necessary to enumerate and evaluate the developmental stages of
the embryos. An attempt was made to classify the celery somatic embryos from nonembryos so that an appropriate time can be decided for the transfer to next culture stage
[7]. Parameters values such as area, length to width ratio, circularity and distance
dispersion were derived from the images of celery cell cultures and subsequently
subjected to train the ANN. After training, the network could not only classify the
globular, heart and torpedo stage embryos and but also successfully predicted the
number of plantlets developed form from heart and torpedo shaped embryos. This is an
example, where ANN could decipher relevant information even from the noisy data.
This work demonstrated an efficient non-destructive approach to identify and classify
the embryogenic cultures on par with human expertise. Such system of classification is
essential for automation, which can economize the process in terms of time and labour.
A pattern recognition system was developed using image analysis system coupled
with ANN classifiers to characterize the somatic embryos of Douglas fir [11].
Geometric features of somatic embryos and their Fourier transformations were
subjected to the neural network based Hierarchical decision tree classification. Normal
embryos were identified with more than 80 percent accuracy. A three layered neural
network topology was used with 19 input nodes representing radius, length, width,
roundness, area, perimeter and their corresponding Fourier coefficients. Hidden layer,
which discriminate the normal and abnormal embryos consisted of 30 nodes, whereas
25 hidden nodes were used to differentiate the developmental stages of the normal.
Back propagation learning algorithm was incorporated into the neural network system
after correlation with the known features. It is apparent from the training phase that the
Fourier features played a major role in distinguishing the normal and abnormal somatic
embryos, whereas size dependent features were the main factor in classifying the
different developmental stages. This pattern recognition system achieved about 85%
accuracy for normal embryos. Thus, it could help in the optimization of developmental
process of somatic embryos. Discarding abnormal embryos could also minimize the low
conversion frequency in the final produce.
3.3. ESTIMATION OF BIOMASS OF PLANT CELL CULTURES
A neural network approach to estimate biomass and sugar consumption rate in cell
cultures of Daucus carota was described by Albiol et al. [8]. The work demonstrated
the relative efficacy of neural network in estimating plant cell mass growth over the
conventional modelling tools. In order to estimate the biomass formation, feed-forward
neural network architecture with bias was employed with one hidden layer. Three
neurons were assigned to the hidden layer in order to achieve lower quadratic error
value for biomass with minimal iterations requirement. There were eight input neurons
for time, biomass, sucrose level, glucose level, fructose level and four output neurons for
58
www.taq.ir
the levels of biomass, sucrose, glucose and fructose. The data for feeding the network
were derived from two different bioreactors with different levels of inoculum and sugar
concentration. A sigmoid function is applied to the neuronal output signal for training
the algorithm. Quadratic error measured from the output of the network was used as an
objective function to change the weights following gradient descent method in a
backward direction. Iterative process is followed for the whole set of inputs until a
convergence criterion is obtained. After the training, new data sets were tested to
validate the performance of the network.
A supervised training was imparted to a three-layered feed-forward network by
correlating the network outputs with the experimental data. During the training phase,
when the data from one bioreactor was used, the network-simulated data pertaining to
both carbohydrate and biomass content correlated poorly with the experimental one.
However, the network predictions were reasonably accurate when trained with two
experiments representing two different culture behaviours. The first experiment was
performed with Biolab reactor with an initial biomass concentration of 0.75 gm/L and
the second one in Colligen bioreactor with a higher inoculum of 0.96 gm/L. Additional
input in the learning process considerably improved the performance. In the validation
step, changes in sugar and biomass evolutions were correctly predicted by the network
output. The method successfully measures the sugar and biomass levels online of plant
cell cultures. The performance of the network was compared with the Extended Kalman
Filter (EKF) approach [13] based on the use of a deterministic mathematical model
(Figure 6). EKF was found to be dependent on several experiments, whereas the
network was able to describe the culture behaviour after training with just two
experiments. Thus, the network approach offers an efficient alternative even with little
experimentation and minimum available information.
3.4. SIMULATION OF TEMPERATURE DISTRIBUTION INSIDE A PLANT
CULTURE VESSEL
Control of microenvironment inside the plant culture vessel is critical for plant growth
[14]. Environmental control such as CO2 concentration, ventilation rate, light intensity,
air temperature inside the culture vessel affects the growth of the regenerated plants. In
particular, increase in air temperature due to high light intensity inhibited the growth.
Controlled cooling of culture vessel has been recommended to reduce the air
temperature and it requires extensive experimentations by varying the factors like:
shape and /or size of the vessel, ambient temperature, head load from light, material of
the container, velocity of blowing air and bottom cooling temperature of culture vessel.
An effective method to determine the forced connective heat transfer coefficient over
the plant culture vessel was developed using a finite element neural network inverse
technique [9] (see also the chapter of Murase et al. in this book).
A finite element model may predict the temperature distribution inside the culture
vessel for which the constants of Nusselt number equation are required. These constant
values were determined through a Kalman filter neural network rout from measured
temperatures of the experiments with the hidden layer comprising of 12 neurons. Four
input neurons were incorporated corresponding to the node temperatures as described in
the finite element model. The simulated temperature values were then fed into a three-
59
www.taq.ir
layered neural network in an iterative manner for adjusting the weights until a
satisfactory learning level has been achieved. The four centre nodal temperatures (of gel
and three air temperatures at three different heights of the culture vessel) were measured
using copper-constantan thermocouples and were approximated by a system of finite
elements. The temperatures at different air velocities were measured and processed
through neural network to estimate the constants of Nusselt equations. Then with these
coefficients, convective heat transfer over the culture vessel surface at different air
velocities was calculated. The errors for air and gel temperatures between experimental
and simulated values were below 5% for air velocities of 1, 2 and 4 ms-1 . The training
data for the neural network were generated by the finite element model from random
values of Nusselt equation constants. The random inputs to the network covered the
entire possible combination of coefficients of convective heat transfers.
Figure 6. Stepwise procedure for estimation of plant cell culture biomass by Kalman filter
approach and neural network approaches. Reprinted with permission from Prof. Manel
Poch, Universitat Autonoma de Barcelona, Spain. [8].
The data is transferred through the neural network and finite element model in a
circulatory fashion. Training the network, the constants of Nusselt equations were
directly and accurately determined by measured temperatures from the experiments.
The generalization feature of the neural network allowed the random inputs to cover the
entire range of convective heat transfer coefficients pertaining to possible temperature
distributions. Training of the network with finite element model outputs, made the
temperature distribution estimation easy and accurate.
60
www.taq.ir
3.5. ESTIMATION OF LENGTH OF IN VITRO SHOOTS
Neural network aided estimation of shoot length of in vitro regenerated rice was
demonstrated by Honda et al. [10]. Digitized images of the regenerated cultures were
captured using CCD camera and fed into computer for data extraction. To assess an
appropriate model for shoot region identification both multiple regression analysis
(MRA) and fuzzy neural network models (FNN-A and B) were studied on comparative
basis. MRA consisted of three different equations and the normalized brightness values
for RGB regions were input into each equation. The outputs of these equations were
positively correlated to the experimental RBG brightness values, which ascertain the
identification of shoot, callus and medium regions for that particular pixel input data
set.
In neural network approach, two different types of FNN were used to distinguish
shoot regions. FNN-A comprised of one model with three inputs and three outputs,
whereas FNN-B consisted of three independent models with three input units and one
output unit per model. In this approach, numerical input values were fuzzified. The
individual nodes of the fuzzy neural hold a sigmoid activation function and the
networks were trained in supervised manner with back propagation algorithm. The
connection weights of the trained model were entered in the colour rule table and
compared with each other to derive the relevance of colour (s) in the model to
distinguish the shoot, callus and medium regions. The extent of complexity in the
relationship between the individual colour components was numerically derived from
the connection weights of the trained neural network. Therefore, the fuzzy neural
network model appears to have a higher level of accuracy in identification of shoots.
Using FNN the shoot recognition was 95% accurate.
Since, FNN-B model was found to be more effective for recognizing callus region
than FNN-A, a trinary image was reconstructed using the outputs of FNN-B model.
This trinary image was then subjected to a two-step method of thinning and extraction
of the longest path based on Hilditch’s algorithm and Tanaka’s algorithms respectively
to separate the shoot region form the rest of the image and estimate its length. The
elongated shoots of the regenerated rice calluses were measured after straightening and
compared with network-simulated values. The average error of only 1.3 mm was
observed between the predicted and actual lengths.
3.6. CLUSTERING OF IN VITRO REGENERATED PLANTLETS INTO GROUPS
One of the prime concerns of in vitro plant micropropagation is the poor survival of
regenerated plants upon ex vitro transfer. The intrinsic quality of the regenerated plants
is largely responsible for its survival during the period of acclimation. Variations are
reflected in the physiological status and in vitro behavioural aspects of the plantlets viz.,
rooting ability, hyperhydric status and adaptability to ex vitro condition etc. These kinds
of variation are not similar to that of well documented aspects of somaclonal variation,
but deserve attention for successful ex vitro transfer.
Development of automatic decision making entity reflecting the variations of in
vitro regenerated plants is necessary to ensure high rate of survival upon ex vitro
transfer. The decision-making may be made in the form of grouping or clustering of
regenerated plants based on their inherent properties. Such decision-making system
61
www.taq.ir
coupled with robotics can results in mechanization of commercial mass propagation.
Since the physiological and behavioural variations among the regenerated plants are
difficult to be resolved by human visual evaluation, machine-vision coupled neural
network based clustering might be an efficient alternative.
For automated clustering of regenerated plants, reliable and contributory features
need to be obtained from the plants, which would help in decision-making. Colour
information based machine vision analysis (MVA) has been acclaimed as a rapid,
sensitive and non-invasive method for qualitative evaluation and quantification of in
vitro regenerated plant cultures [15]. It has been suggested that the photometric
parameters could serve as reliable indicators for assessing the behaviour of regenerable
cultures. Leaf spectral reflectance and brightness intensity can be captured as digitized
images for compilation of input features which can further be processed with neural
network algorithm to interpret and project the inherent variations. In this way, a
functional activity in a biological system can be correlated to the minute machineobserved colour based information.
We test the hypothesis that whether regenerated plants can be sorted out into groups
based on their photometric behaviour using image analysis system coupled with neural
network algorithm. It is well understood that the successful clustering of regenerated
plants gives an opportunity to identify and select plants amenable for ex vitro survival.
A neural network based image processing method was developed for clustering of
regenerated plantlets of gladiolus based on the leaf feature attributes in Red, Blue and
Green colour regimes [12].
The main objective of any clustering model would be to find a valid organization of
the data with respect to the inherent structure and relationship among the inputs. ART2
network, originally developed by Carpenter and Grossberg [16], is one such model
which is configured to recognize invariant properties within the given problem domain.
From the luminosity and trichromatic components of the leaf images, 12 attributes per
individual plantlets were extracted. These 12 attributes constituted the input pattern for
a single plantlet and were fed to ART2 algorithm, which was compiled by ‘C’
programming. Unlike ART1, ART2 model has the distinct ability to process the leaf
input patterns, which are analogue-valued.
The description that follows is intended to outline the generalized ART2 network
principles. ART2 network is divided into two subsystems namely attentional subsystem
and orienting subsystem. The basic function of attentional subsystem is to establish valid
categories based on salient features of the input patterns. The attentional subsystem
forms a platform for establishing resonance conditions between activity patterns flowing
in feed-forward and feed -back direction. When such bottom-up input pattern is found
superposable to the top-down expectation pattern, it is regarded as a constituent of that
established category. The attentional subsystem is comprised of F0, F1 and F2 layers. F0
layer comprises of 4 sub-layers namely, wio xio vio uio and F1 layer contains 6 sub-layers,
namely wio xio vio uio pio qio. The input nodes contain a nonlinear transfer function with a
threshold value (ș). The noise level in the input information dictates the nodal
activation. F0 and F1 layers of the attentional subsystem function in order to enhance
significant aspects of the input signals. This is particularly necessary for analogue input
patterns since the difference between the possible values of a feature is much smaller
than the difference that is generally described in terms of binary values.
62
www.taq.ir
The parametric conditions laid down for the basic ART2 clustering analysis are as
follows,
a > 0; b > 0 ; d = 0 to 1; c such that c X d / (1-d) is 1 ; e <<1; ș d 1
‘a’ and ‘b’ are the model gain parameters. These parameters influence the stability of
the network. Lower values of à’ and `b’, allow wider range of vigilance parameter
values to be used and also consequently results in the formation of increasing number of
stable categories even when trained with fewer number of learning data sets. However,
it must be noted that higher values of à’ and `b’ could ultimately result in one pattern
getting allocated to more than one category. The parameters `c’ and `d’ are valued as
per the original ART2 model where their relationship is pre-established. The primary
function of parameter ‘e’ is to prevent a divide by zero condition. Therefore, its value is
kept relatively very small.
Figure 7. Block diagram of entities in ART2 network.
The values that are assigned to the network parameters in our venture are as follows:
a = 10; b = 10; c = 0.1; d = 0.8; e = 0.000001 and ș = 0.0001
The activities in F0 and F1 layer and the direction of the flow of signals are depicted in
Figure 7. In F0 and F1 layers, the raw input values are normalised. The activity function
at F0 and F1 layers is defined by the following condition,
63
www.taq.ir
xifx t T
f x ®
¯0
(1)
Where, T is the threshold value in non-linear function with a positive constant of less
than unity. The output of F0 layer (uio) forms the input to F1 layer. F2 layer sums-up
processed input activity pattern (pi) after the normalization of input pattern. The node
that has maximum summation value is considered as th winning output category node.
In the first cycle, since the top-down weights (Zji) are assigned as zero, a random
selection determines the winning output node. During such random selection, the initial
values of the bottom-up connection weight (Zij) from ‘i’ input node towards ‘j’ output
node is given by,
zij
1
1 d M
(2)
Where, d is the model parameter whose values are between 0 and 1 and M is the
dimension of the supplied input patterns. When resonance condition between the
bottom-up and top-down expectation pattern is insufficient to overcome the threshold
set by the VP (ȡ), there will be removal of winning node by a reset vector (r). Then a
new parallel search cycle is carried-out until a winning node is selected that brings
about resonance surpassing the threshold. When that happens, the adaptive weights
associated with winning F2 node are updated accordingly. The learning equations for
bottom-up and top-down adaptive weights connecting F1 and F2 layers are calculated
considering the following condition,
dT j maxT j & j is not reseted
g yi ®
¯0 Otherwise
(3)
In the matching process, the two F1 sub-layers that take part are ‘pi’ and ‘ui’. During
learning, the activity of the units on the ‘pi’ layer changes as top-down weights changes
on the ‘pi’ layer. The ‘ui’ layer remains stable during training, therefore including it in
the matching process prevents the occurrence of reset while learning of a new pattern is
underway. The reset vector (r) situated in the orienting subsystem determines the degree
of match between short term memory pattern at F1 layer and long term memory pattern
at F2 layer. This reset vector is calculated after all the F1 layers have been updated to
reflect the effects of feed-back from F2 layer. If reset value is higher than the VP value
then the winning node is retained as an established matching category and on the
contrary, if the reset value is lower than VP then the winning node is disabled
accordingly.
In our study the number of generated groups increased from 1 to 2 with the VP
range over 0.985. The network validity was proved when the class separability was
retained with another similar set of test input patterns. Leaves having maximum
similarity in terms of inherent pixel properties fall in a particular group. Hence, it has
64
www.taq.ir
been demonstrated that the leaf photometric property could provide a classifying feature
with which the discrepancies among the regenerated plantlets can be projected. The use
of flatbed scanning machine instead of CCD camera, ‘C’ program based compilation of
the ART-algorithm in a PC with 1.6 GHz clock speed and 256 MB random access
memory in lieu of professional ready made off the shelf software rendered the whole
process right from the image acquisition to analysis, cost effective. The component
steps of the image analysis systems are presented in Figure 8. Such an approach may
provide a means of reliable and objective measurement for selecting plants amenable
for ex vitro survival and quality control in commercial micropropagation.
Figure 8. Component steps of machine vision analysis for sorting of in vitro regenerated
plants into groups. Adapted from Mahendra et al. (2004) [12].
4. Conclusions and future prospects
The use of ANN is increasingly becoming most preferred methodology to model the
complex biological responses. ANNs can also play central role as highly potential
predictive modelling tool in in vitro plant culture studies. Neural computing offers
reliable and realistic approach for describing in vitro culture of plant species even with
minimal available information. The successes obtained after applying neural network
technology have been phenomenal with a relatively modest experimental effort while
65
www.taq.ir
consuming minimum amount of time. Maximum inference has been derived from
relatively simplistic experimental procedures. The ability of the ANN to accurately
simulate even under altered conditions could be highly encouraging in design of
cultivation systems on large scale. Various image processing methods have been
developed successfully for assessing culture types, biomass production etc. but in order
to bring them to a usable form, neural network solutions offer attractive incentives.
ANN can be modulated to simulate the metabolism of the in vitro plants under a
given set of conditions. It could be useful in estimating the amount of secondary
metabolites that could accumulate at a specified time period and also the time at which
one can derive maximum yield. ANN based prediction of the behaviour of the in vitro
derived plants in terms of their ex vitro survival rate and their rooting or organogenic
ability could also be useful in large scale propagation. The outcome of the neural
computations can be directed to mechanize systems to automate online processing of
plant cell cultures, sub-culturing and quality based segregation of plant tissues all in
aseptic fashion.
Acknowledgement
Financial assistance to VSS Prasad from CSIR, New Delhi as a SRF is acknowledged.
References
[1] Nazmul Karim, M.; Yoshida, T.; Rivera, S. L.; Saucedo, V. M.; Eikens, B. and Oh, G. S. (1997) Global
and local neural network models in biotechnology: Application to different cultivation processes. J.
Ferment. Bioengg. 83: 1-11.
[2] Hashimota, Y. (1997) Applications of artificial neural networks and genetic algorithms to agricultural
systems. Comput. Electro. Agri. 18: 71-72.
[3] Patnaik, P. R. (1999) Applications of neural networks to recovery of biological products. Biotechnol. Adv.
17: 477-488.
[4] Hudson, D. L. and Cohen, M. E. (Eds.) (2000) Neural networks and artificial intelligence for biomedical
engineering. The Institute of Electric and Electronics Engineers Press Inc., New York.
[5] Haykin, S. (1994) Neural networks: A comprehensive foundation. Macmillan College Publishing Co.,
New York.
[6] Tani, A.; Murase, H.; Kiyota, M. and Honami, N (1992) Growth simulation of alfalfa cuttings in vitro by
kalman filter neural network. International Symposium on Transplant Production Systems. Acta. Hort.
319.
[7] Uozumia, N; Yoshinoa, T.; Shiotanib, S.; Sueharaa, K. I.; Araib, F.; Fukudab, T. and Kobayashi, T.
(1993) Application of image analysis with neural network for plant somatic embryo culture. J. Ferment.
Bioengg. 76: 505-509.
[8] Albiol, J.; Campmajo, C.; Casas, C. and Poch, M. (1995) Biomass estimation in plant cell cultures: A
neural network approach. Biotechnol. Prog. 11: 8-92.
[9] Suroso; Murase, H.; Tani, A.; Hoami, N.; Takigawa, H. and Nishiura, Y. (1996) Inverse technique for
analysis of convective heat transfer over the surface of plant culture vessel. Trans. ASAE. 39: 2277-2282.
[10] Honda, H.; Takikawa, N.; Noguchi, H.; Hanai, T. and Kobayashi, T. (1997) Image analysis associated
with fuzzy neural network and estimation of shoot length of regenerated rice callus. J. Ferment. Bioeng.
84: 342-347.
[11] Zhang, C.; Timmis, R. and Shou Hu, W. (1999) A neural network based pattern recognition system for
somatic embryos of Douglas fir. Plant Cell Tissue Org. Cult. 56: 25-35.
[12] Mahendra; Prasad, V. S. S. and Dutta Gupta, S. (2004) Trichromatic sorting of in vitro regenerated plants
of gladiolus using adaptive resonance theory. Curr. Sci. 87: 348-353.
66
www.taq.ir
[13] Albiol, J.; Robuste, J.; Casas, C. and Poch, M. (1993) Biomass estimation in plant cell cultures using an
extended kalman filter. Biotechnol. Prog. 9: 174-178.
[14] Morohoshi, N. and Komamine, A. (Eds.) (2001) Molecular Breeding of Woody Plants. Elsevier Sci. B.
V., The Netherlands.
[15] Honda, H.; Ito, T.;Yamada, J;Hanai, T.;Matsuoka, M. and Kobayashi, T. (1999) Selection of
embryogenic sugarcane callus by image analysis. J. Biosci. Bioeng. 87: 700-702.
[16] Carpenter, G. A. and Grossberg, S. (1987) ART2: Self organisation of stable category recognition codes
for analogue input patterns. Appl. Optics. 26: 4919-4930.
67
www.taq.ir
EVALUATION OF PLANT SUSPENSION CULTURES BY TEXTURE
ANALYSIS
YASUOMI IBARAKI
Department of Biological Science, Yamaguchi University, Yoshida 16771, Yamaguchi-shi, Yamaguchi 753-8515, Japan - Fax: 81-83-933-5864 Email: [email protected]
1. Introduction
Plant cell suspension culture has been widely used as a way for cell proliferation in
research and is extending to commercial use. To make the best use of this technique, it
is essential to maintain cell quality. Selection of cell suspensions having desirable
properties is a routine work in plant cell suspension culture [1]. Image analysis
techniques appear to be one of the promising methods for evaluation of cell suspension
cultures because it can offer non-destructive monitoring of culture giving an objective
index for visual information [1,2]. The macroscopic visual appearance of cell
suspensions may vary with colour and size distribution of cell aggregates in the cell
suspensions, depending on culture conditions, culture periods, or cell lines. Hence, the
visual texture of a macroscopic image of a cell suspension may be used for evaluation
of cultured cell quality [1,3].
In this chapter, the feasibility and problems of methods for the non-destructive
evaluation of cell suspension cultures will be discussed, focusing on texture analysis of
macroscopic images of cell suspensions. First, macroscopic images will be compared
with microscopic images from the viewpoint of their use for non-destructive evaluation
of cell suspension cultures, and basics of texture analysis for biological objects will be
explicated. Next, as an example of application of texture analysis for macroscopic
images, a research on evaluation of somatic embryogenic potential of carrot cell
suspension culture will be introduced.
2. Microscopic and macroscopic image uses in plant cell suspension culture
Normally, objects in cell suspension culture are single cells or cell aggregates.
Therefore, to identify cells or cell aggregates, images of cell suspensions acquired using
microscopy, are necessary. As plant cells are normally several micrometers to several
tens of micrometers in size, a spatial resolution of at least several micrometers per pixel
is needed in microscopic images to analyze single cells or small cell aggregates. Use of
microscopic images has the advantage of allowing direct observation of individual cells,
69
© 2008 Springer.
www.taq.ir
www.taq.ir
Y. Ibaraki
cell aggregates and differentiated cell masses. However, this microscopic image
analysis has difficulties in image acquisition [1]. Generally, to acquire microscopic
images, sampling of the culture is necessary. Sampling may be destructive with risks of
contamination, and is labour-intensive. In addition, sampling raises questions of
whether the sample population is truly representative of the cell suspension, and it may
be necessary to increase the number of samples or use effective statistical methods [1].
By using an inverted microscope attached with a camera or a long working distance
microscopic CCD camera, image can be acquired without sampling. However, it is
difficult to obtain microscopic images of suspended cells suitable for direct observation
of individual cells and cell aggregates because of cell overlapping by sedimentation or
limitation in working distance. In addition, whether the populations recorded in sampled
images are truly representative remains a problem.
Several microscopic imaging system in which an image of suspended cells is
acquired in an imaging cell connected to a bioreactor, have been proposed. Grand
d’Esnon et al. [4] first reported this type of system for acquiring cell microscopic
images. Suspended Ipomoea batatas Poir. cells were passed into the imaging cell by a
peristaltic pump from the bioreactor. This system was used to monitor the population
dynamics of embryogenic and non-embryogenic cell aggregates in cell suspension
cultures used for somatic embryo production. Smith et al. [5] have developed a similar
system that evaluated pigment production of Ajuga reptans cells. Ibaraki et al. [6] also
developed a system to acquire images of carrot somatic embryos (Daucus carota L.) for
sorting. Harrell et al. [7] developed an improved system and measured cell aggregate
distribution and growth rate in embryogenic cell suspension cultures of Ipomoea
batatas Lam. In this system, to avoid cell damage the cell aggregates could not be
allowed to go through the pumping unit, and a method to calculate total reactor
population from the number of observed aggregates was proposed. These methods are
effective for serial quality evaluation in cell suspension cultures. However, it should be
noted that the population density of single cells and cell aggregates is crucial if image
analysis is used to measure the properties of individual cells and cell aggregates. Low
cell population density is needed to prevent cells from overlapping, and this may not be
optimal for cell growth or metabolite production [1].
In contrast, macroscopic images have an advantage in imaging and have been used
for quality evaluation of cell suspensions although applications are limited to a few
studies. A macroscopic image of a cell culture is defined as an image viewed with
normal or macro lens whose field of view contains almost a whole culture [1].
Macroscopic images can be acquired from the outside of a culture vessel without special
devices if the culture vessel has transparent walls, i.e., it is perfectly non-destructive
imaging. Depending on the imaging devices, these images have spatial resolutions of
several hundreds of micrometers per pixel and do not allow us to identify a small cell
aggregate. However, macroscopic images have often been used for quantification of cell
masses on solid media [8,9,10] and in cell suspensions [11] because they included one
whole culture in their fields of view. Moreover, colour /grey level analysis and/or
texture analysis of macroscopic images of suspension cultures can provide us with
information related to status of suspended cells and tissues. Texture analysis has the
potential of characterizing individual objects in a macroscopic image, in which the
individual objects were not clearly identified [12]. Experimental evaluations in plant
70
www.taq.ir
Evaluation of plant suspension cultures by texture analysis
cell culture very frequently include visual examinations [2]. Image analysis of a
macroscopic culture image may substitute for the visual examination, supporting
objective decision and contributing to improvement in reproducibility in plant cell
culture.
3. Texture analysis for macroscopic images of cell suspensions
3.1. TEXTURE FEATURES
As simple texture features, mean grey level, variance, range (i.e., the difference between
maximum and minimum values of grey level), and other statistical features derived
from grey level histogram such as skewness and kurtosis, are used for classification and
segmentation of images based on texture although these texture features can not involve
information on spatial distribution.
Texture analysis methods considering spatial distribution include two-dimensional
frequency transformation, grey level run lengths method, spatial grey level dependence
method, etc. Two-dimensional frequency transformation method has been widely used
for image analysis. It can derive the power spectrum image (frequency-domain image),
which expresses periodic features in the image texture. From power spectrum images,
wedge-shaped features related to texture direction and ring-shaped features expressing
periodic characteristics can be extracted.
In grey level run lengths method [13], features are extracted from the matrix which
is a set of probabilities that a particular-length line consisting of pixels with the same
grey level will occur at a distinct orientation. It is valid for analysis of band pattern
texture.
Texture features extracted using spatial grey level dependence method (SGDM)
developed by Haralick et al. [14] have been often used for texture analysis for
biological objects. In SGDM, a co-occurrence matrix is determined and 14 texture
features are calculated from the matrix. The co-occurrence matrix is a set of the
probabilities P(i,j) that a combination of a pixel at one particular grey level (i) and
another pixel at a second particular grey level (j) will occur at a distinct distance (d) and
orientation (ș) from each other. Of the 14 features, major features are as follows:
N 1 N 1
Angular Second Moment
¦¦ P(i, j)
2
(1)
i 0 j 0
N 1
Contrast
¦ n ¦ P (i , j )
2
n 0
(2)
|i j| n
71
www.taq.ir
Y. Ibaraki
N 1 N 1
¦¦ ijp(i, j) - P
Correlation
xPy
i 0 j 0
(3)
V xV y
N 1 N 1
Entropy
¦¦ p(i, j ) log( p(i, j))
(4)
i 0 j 0
Where, N is the number of grey levels, and µx, µy, ıx, ıy denote the mean and standard
deviation of the row and column sums of the co-occurrence matrix, respectively.
Briefly, “Angular Second Moment” is a measure of homogeneity, “Contrast” is a
measure of local contrast, “Entropy” is a measure of the complexity or randomness of
the image, and “Correlation” is a measure of grey-tone liner-dependencies. The number
of grey levels, N, is often lessened for reducing calculation time and for suppressing
noise effect. If the image is assumed to be isotropic, only one orientation (ș) is often
tested. Moreover, recently, texture analysis using the colour co-occurrence matrix has
been used [15].
A wide variety of new texture analysis methods have been proposed extensively in
various research fields. Tuceryan and Jain [16] divided texture analysis methods into
four categories: statistical, geometrical, model-based, and signal processing. Of these
categories, histogram-derived features, grey level run lengths method, and SGDM are
classified into statistical methods, and two-dimensional frequency transformation is
classified into signal processing methods. Geometrical methods consider texture to be
composed of texture primitives, attempting to describe the primitives and the rules
governing their spatial organization [17]. Model-based methods hypothesize the
underlying texture process, constructing a parametric generative model, which could
have created the observed intensity distribution [17].
3.2. TEXTURE ANALYSIS FOR BIOLOGICAL OBJECTS
In remote sensing, texture analysis has been used for classification of land use or plant
species identification extensively. In proximal remote-sensing for plant canopies,
applications of texture analysis have been also reported. Shearer and Holmes [15]
identified plant species using colour co-occurrence matrices, which were derived from
image matrices for each colour attribute: intensity, hue, and saturation. Shono et al. [12]
compared the effectiveness of several methods for texture analysis, including grey level
run lengths method, SGDM, and power spectrum method, on estimation of the species
composition in the pasture filed.
In addition, in the filed of quality evaluation in agriculture, machine vision systems
based on texture features have been used. Sayeed et al. [18] evaluated snack quality by
neural network using textural and morphological features. Maturity in shell-stocked
peanuts was detected by the histogram characteristics or the texture descriptor derived
from the analysis of gradient images [19]. Texture analysis which is based on the
frequency of co-occurrence of a random event and is named as Frequency Histogram of
72
www.taq.ir
Connected Elements was used for detection and recognition of cracks in wood boards
[20]. Shono [21] analyzed leaf orientation by texture features extracted by power
spectrum method. Murase et al. [22] quantified plant growth by analyzing texture
features using neural network.
Texture analysis has been used for biological objects besides plants extensively. The
applications include assessment of chromatin organization in the nucleus of the living
cell [23], and medical applications for brain MR images [24], for bone radiographs [25],
and for pulmonary disease images [26].
3.3. TEXTURE ANALYSIS FOR CELL SUSPENSION CULTURE
Although applications of texture analysis for plant cell suspension culture are still
limited to a few studies, texture analysis has the potential of evaluating and/or selecting
cell suspension cultures. The macroscopic visual appearance of cell suspensions reflects
on colour and size distribution of cell aggregates, which may be indicators of cell
suspension culture status. Cell aggregate size distribution patterns in cell suspension
culture vary significantly between cell lines and also a consequence of culture age and
culture conditions [27,28]. It has been reported that the visual appearance of suspension
cultures changes with the number of subcultures [29] or with variations in embryogenic
potential [3,29]. In fact, statistical texture features were effective for describing the
difference in macroscopic appearances between carrot embryogenic and nonembryogenic suspensions [3]. The study will be introduced in 4.2. Texture analysis is
expected to contribute to maintenance of cell quality in plant suspension culture,
offering objective index for macroscopic appearance of suspension culture.
3.4. CONSIDERATIONS FOR APPLICATION OF TEXTURE ANALYSIS
It should be noted that as texture features are not the direct measures of biological
properties in many cases, it is required to determine the relationships between texture
features and the targeted biological properties by modelling methods such as regression
analysis [3] and artificial neural network [18,20,22] to use the features for evaluation of
biological properties. In addition, dependency of texture features on the experimental
set-up including image acquisition, sampling, and pre-processing, should be considered
[17]. All experimental results should be considered to be applicable only to the reported
set-up [17]. For routine use of texture analysis of macroscopic images, simple indices
for describing cell suspension culture properties without the complicated model are
required. In addition, more efforts for developing the robust way to acquire a
macroscopic image of a cell suspension should be made in view of dependency of
texture features on image acquisition set-up.
4. Evaluation of embryogenic potential of cultures by texture analysis
4.1. EVALUATION OF EMBRYOGENIC POTENTIAL OF CULTURES
The productivity of somatic embryos depends on the quality of embryogenic cultures
[3]. The embryogenic potential of cultures must be sustained in maintenance phase for
73
www.taq.ir
Y. Ibaraki
the stable production of somatic embryos. The embryogenic potential depends on
genotypes. Moreover, it can change with culture period and is affected by medium
composition and environmental conditions. To monitor embryogenic potential of
culture would be useful to stably produce somatic embryos [30].
Using microscopic observation, a pro-embryogenic mass (PEM), which is a cell
cluster to become somatic embryos under certain conditions, could be identified. In a
number of systems studied to date, PEMs shared similar structural features. They
consist of small and highly cytoplasmic cells which often have an accumulation of
starch within the plastids [31]. On the other hand, non-embryogenic cells are large and
vacuolated. Therefore, a PEM could be selected with regard to its transparency and
shape under microscopy. The amount of PEMs in cell suspensions may be one direct
index for determining the embryogenic potential of the culture. In a similar way, the
amount in cultures of other embryogenic tissues as materials for embryo production
such as embryo suspensor masses and early globular embryos can be used for
evaluation of cultures.
Microscopic image analysis for suspension culture could be used to select PEMs.
Grand d’Esnon et al. [4] monitored population dynamics of PEMs in suspension
cultures of Ipomoea batatas for somatic embryo production using image analysis. PEMs
and non-embryogenic cell aggregates were divided by using a correlation between the
size and the mean transparency of the object.
Culture growth rate may be one of indices for evaluation of the embryogenic
potential [1]. Differences in growth characteristics between embryogenic and nonembryogenic cultures have been reported in maize suspension culture [28], in carrot
suspension culture [11,32], and in Ipomoea batatas callus culture [33]. Growth rates can
be calculated through non-destructive cell quantification by image analysis. There have
been several reports on image-analysis-based quantification of cells on gelled media
[8,9,10,34]. In addition, Ibaraki and Kurata [11] quantified embryogenic suspension
cultures by image analysis of macroscopic images of the suspensions. They showed the
relationship between growth rate estimated by image analysis and embyrogenic
potential of carrot embryogenic culture.
4.2. TEXTURE ANALYSIS BASED EVALUATION OF EMBRYOGENIC
POTENTIAL
Other indices to be potentially used for evaluation of suspension culture are colour, cell
aggregate distribution, and consequent macroscopic texture [1]. Ibaraki et al. [3]
reported the system for evaluation of embryogenic potential of cell suspension cultures
based on texture analysis. They acquired macroscopic images of carrot cell suspensions
from the bottom of a culture vessel (Erlenmeyer flask) with a video camera (GR-S95,
JVC) using transmitted light. The video signal was digitized as a 24-bit RGB colour
image whose size was 640 by 480 pixels. As the B component of the RGB was more
sensitive to yellow carrot cells than the other two components, each image was
converted into an 8-bit monochrome image based on the B value. A part of the flask
bottom in the image was extracted as an elliptic region and transformed into a circle
with 400-pixel diameter (Figure 1). In this condition, the spatial resolution in the image
was about 0.23 mm/pixel. Texture features were extracted using SGDM. Of 14 features
in SGDM, 3 features, Angular Second Moment, Contrast, and Entropy were calculated
74
www.taq.ir
from co-occurrence matrix and tested. Actual embryogenic potential of a cell suspension
was determined by the number of PEMs in the unit volume suspension (hereafter, PEM
density) or total number of embryos induced using each cell suspension.
Figure 1. Macroscopic images of carrot cell suspension viewed form the bottom of culture
vessel. A part of the flask bottom in the original colour image (A) was extracted after
conversion into 8-bit monochrome image based on the B component value as an elliptic
region and transformed into a circle with 400-pixel diameter (B).
Different carrot cell suspensions had various embryogenic potentials expressed by the
PEM density. Differences in visual appearance due to differences in cell aggregate size
distribution pattern between embryogenic and non-embryogenic suspensions were
observed (Figure 2). Images of cell suspensions possessing high embryogenic potential
had course texture, while those of non-embryogenic suspension had fine texture. In
embryogenic cell suspensions, many large cell aggregates could be observed. In contrast
to this, non-embryogenic suspensions had few large cell aggregates and consisted mainly
of small cell aggregates. Several reports have been shown difference in cell aggregate
size distribution patterns between embryogenic and non-embryogenic cultures [28,35].
The difference in textural appearance due to cell aggregate distribution patterns could
be detected by texture analysis. The most useful texture feature for evaluating the
embryogenic potential was Entropy, which is a measure of complexity of an image.
Images of cell suspensions with higher PEM density had higher values of texture feature
Entropy (Figure 3). In addition, suspensions with higher values of texture feature
Entropy have the potential to produce more somatic embryos (Figure 4). These results
suggested that texture analysis of a macroscopic image of a cell suspension could be
used to evaluate the embryogenic potential of the suspension.
75
www.taq.ir
Y. Ibaraki
Figure 2. Images of embryogenic and non-embryogenic suspensions.
Figure 3. Relationship between texture feature entropy and PEM density when the number
of grey level =8 (n=43). Reprinted from Ibaraki et al. (1998) [3].
76
www.taq.ir
Figure 4. Relationship between texture feature entropy when the number of grey level =8
and number of induced somatic embryos. Reprinted from Ibaraki et al. (1998) [3].
Image analysis has potential to provide simple, non-destructive, and objective quality
evaluation of cultured cells for plant cell suspension culture. As compared with
microscopic images, macroscopic images are more easily acquired without sampling,
showing the potential for non-destructive evaluation. The visual texture of a
macroscopic image of a cell suspension can be an indicator of cultured cell quality. The
texture analysis of the macroscopic image was used for evaluation of embryogenic
potential in cell suspension cultures. Texture analysis techniques are expected to
contribute to maintenance of cell quality in plant cell suspension culture. Texture
analysis is now used extensively for biological objects in various areas and novel
methods have been reported. These technologies are expected to be transferred to plant
tissue culture area.
References
[1] Ibaraki, Y. and Kurata, K. (2000) Application of image analysis to plant cell suspension cultures. Compu.
Electron. Agri. 30:193-203.
[2] Smith, M.A.L. (1995) Image analysis for plant tissue culture and micropropagtion. In: Aitken-Christie, J.;
Kozai, T. and Smith, M.A.L. (Eds.) Automation and Environmental Control in Plant Tissue Cultures.
Kluwer Academic Publishers, Dordrecht, The Netherlands; pp. 145-163.
77
www.taq.ir
Y. Ibaraki
[3] Ibaraki, Y.; Kaneko, Y. and Kurata, K. (1998) Evaluation of embryogenic potential of cell suspension
culture by texture analysis. Trans. ASAE 41: 247-252.
[4] Grand d'Esnon, A.; Chee, R.; Harrell, R.C. and Cantliffe, D. J. (1989) Qualitative and quantitative
evaluation of liquid tissue cultures by artificial vision. Biofutur 76:S3.
[5] Smith, M.A.L.; Reid, J.F.; Hansen, A.C.; Li, Z. and Madhavi, D.L. (1995) Non-destructive machine vision
analysis of pigment-producing cell cultures. J. Biotechnol. 40:1-11.
[6] Ibaraki, Y.; Fukakusa, M. and Kurata, K. (1995) SOMES2: Image-analysis-based somatic embryo sorter.
Current Plant Science and Biotechnology in Agriculture 22: 675-680.
[7] Harrell, R.C.; Bieniek, M. and Cantiffe, D.J. (1992) Non-invasive evaluation of somatic embryogenesis.
Biotechnol. Bioeng. 39: texture analysis 378-383.
[8] Smith, M.A.L. and Spomer, L.A. (1987) Direct quantification of in vitro cell growth through image
analysis. In Vitro Cell. Dev. Biol.-Plant 23: 67-74.
[9] Olofsdotter, M. (1993) Image processing: a non-destructive methods for measuring growth in cell and
tissue culture. Plant Cell Rep. 12: 216-219.
[10] Anthony, P.; Davey, M.R.; Power, J.B.; Washington, C. and Lowe, K.C. (1994) Image analysis
assessments of perfluorocarbon- and surfactant- enhanced protoplast division. Plant Cell Tissue Org.
Cult. 38:39-43.
[11] Ibaraki, Y. and Kurata, K. (1997) Image analysis based quantification of cells in suspension cultures for
producing somatic embryos. Environ. Control Biol. 35: 63-70.
[12] Shono, H.; Okada, M. and Higuchi, S. (1994) Texture analysis of photographic images from close
distance: An application to estimate species composition in a mixed pasture field (in Japanese with
English abstract). J. Agri. Meteorol. 49: 227-235.
[13] Galloway, M.M. (1975) Texture analysis using grey level run lengths. Computer Graphics Image
Processing 4: 172-179.
[14] Haralick, R. M.; Shanmugam, K. and Dinstein, I. (1973) Textural features for imaging classification.
IEEE Trans. Sys. Man Cybernet. SMC-3: 610-621.
[15] Shearer, S.A. and Holmes, R.G. (1990) Plant identification using colour co-occurrence matrixes. Trans.
ASAE 38: 2037-2044.
[16] Tuceryan, M. and Jain, A.K. (1998) Texture analysis. In: Chen, C.H.; Pau, L.F. and Wang, P.S.P. (Eds.)
The Handbook of Pattern Recognition and Computer Vision. World Scientific Publishing Co.,
Hackensack, NJ; pp. 207-248.
[17] Ojala, T. and Pietikäinen, M. Texture analysis. In: Fisher, R.B. (Ed.) CV online: The evolving,
Distributed,
Non-proprietary,
On-Line
Compendium
of
Computer
Vision
(http://homepages.inf.ed.ac.uk/rbf/CVonline /LOCAL_COPIES/OJALA1/texclas.htm ).
[18] Sayeed, M.S.; Whittaker, A.D. and Kehtarnavaz, N. D. (1995) Snack quality evaluation method based on
image feature and neural network prediction. Trans. ASAE 38: 1239-1245.
[19] Ghate, S.R.; Evans, M.D.; Kvien, C.K. and Rucker K.S. (1993) Maturity detection in peanuts (Arachis
hypogaea L.) using machine vision. Trans. ASAE 36: 1941-1947.
[20] Guisado, M.A.P. and Gómez-Allende, D.M. (2001) Wood texture analysis by combining the connected
elements histogram and artificial neural networks. In: Mira, J. and Prieto, A. (Eds.) Bio-Inspired
Applications of Connectionism-IWANN 2001.Springer-Verlag, Heidelberg; pp.160-167.
[21] Shono, H. (1995) A new method of image measurement of leaf tip angle based on textural feature and a
study of its availability (in Japanese with English abstract). Environ. Control Biol. 33:1970-207.
[22] Murase, H.; Honami, N. and Nishiura, Y. (1994) A neural network estimation technique for plant water
status using textural features of pictorial data of plant canopy. Acta Hort. 339: 255-262.
[23] Rousselle C.; Paillasson, S.; Robert-Nicoud, M. and Ronot, X. (1999) Chromatin texture analysis in
living cells. Histochemical J. 31:63-70.
[24] Zhang, Y.; Zhu, H.; Ferrari, R.; Wei, X.; Eliasziw, M.; Metz, L.M. and Mitchell, R. (2003) Texture
analysis of MR images of minocycline treated MS patients. In: Elli, R.E. and Peters T.M. (Eds.) MICCAI
2003, LNCS 2878. Springer-Verlag, Heidelberg; pp. 786-793.
[25] Lespessailles, E.; Roux, J.P.; Benhamou, C.L.; Arlot, M.E.; Eynard, E.; Harba, R.; Padnou, C. and
Meunier, P.J. (1998) Fractal analysis of bone texture on os calcis radiographs compared with trabecular
microarchitecture analysed by histomorphometry. Calcified Tissue Int. 63: 121-125.
[26] Sutton, R. and Hall, E.L. (1972) Texture measures for automatic classification of pulmonary disease.
IEEE Trans. Comput. C-21: 667-676.
[27] Kieran, P.M.; MacLoughlin, P.F. and Malone, D.M. (1997) Plant cell suspension cultures: some
engineering considerations. J. Biotechnol. 59: 39-52.
78
www.taq.ir
[28] Stirn, S.; Hopstock, A. and Lorz, H. (1994) Bioreactor cultures of embryogenic suspensions of barley
(Hordeum vulgare L.) and maize (Zea mays L.). J. Plant Physiol. 144: 209-214.
[29] Molle, F.; Dupuis, J.M.; Ducos, J.P.; Anselm, A.; Crolus-Savidan, I.; Petiard, Y. and Freyssinet, G.
(1993) In: Redenbaugh, K. (Ed.) Synseeds. CRC press, Boca Raton; pp. 257-287.
[30] Ibaraki, Y. and Kurata, K. (2001) Automation of somatic embryo production. Plant Cell Tissue Org.
Cult. 65: 179-199.
[31] Yeung, E.C. (1995) Structural and developmental patterns in somatic embryogenesis. In: Thorpe, T.A.
(Ed.) In Vitro Embryogenesis in Plants. Kluwer Academic Publishers, Dordrecht, The Netherlands; pp.
205-247.
[32] Smith, S.M. and Street, H.E. (1974) The decline of embryogenic potential as callus and suspension
cultures of carrot (Daucus carota L.) are serially subcultured. Ann. Bot. 38: 223-241.
[33] Zheng, Q.; Dessai, A.P. and Parkash, C.S. (1996) Rapid and repetitive plant regeneration in sweet potato
via somatic embryogenesis. Plant Cell Rep.15: 381-385.
[34] Hirvonen, J. and Ojamo, H. (1988) Visual sensors in tracking tissue growth. Acta Hort. 230: 245-251.
[35] van Boxtel, J. and Berthouly, M. (1996) High frequency somatic embryogenesis from coffee leaves.
79
www.taq.ir
PART 2
BIOREACTOR TECHNOLOGY
www.taq.ir
www.taq.ir
BIOENGINEERING ASPECTS OF BIOREACTOR APPLICATION IN PLANT
PROPAGATION
SHINSAKU TAKAYAMA1 AND MOTOMU AKITA2
Department of Biological Science and Technology, Tokai University, 317
Nishino, Numazu, Shizuoka 410-0315, Japan. – Fax: 81-263-47-1879 –
2
Department of Biotechnological Science, Kinki University, 930
Nishimitani, Uchita, Naga, Wakayama 649-6493, Japan – Fax: 81-73677-4754 – Email: [email protected]
1
1. Introduction
A large number of commercially important plants including important vegetatively
propagated crops such as vegetables, flowers, ornamentals, fruit trees, woody and
medicinal plants, etc., are vegetatively propagated by tissue culture. Tissue culture is
carried out in most of countries in the world, and the number of plants propagated was
600 millions for one year over the world which is the best available estimates as cited in
Altman and Loberant (2000) [1]. The culture technique generally used for commercial
tissue culture propagation is the agar culture which requires large number of small
culture vessels and labour, and results in the requirement of many laminar air flow clean
benches, large autoclave(s), large culture spaces equipped with illuminated shelves,
electric energy, etc. This is the major cause for both limited propagation efficiency and
high production costs.
In order to overcome these problems, large-scale propagation technique with simple
culture protocol with least equipments and reduced production cost should be adopted.
Many attempts for establishing large-scale production of propagules with simple
production facilities and techniques have been made including robotics,
photoautotrophic cultures, bioreactor techniques, etc. [2]. Among them, bioreactor
technique seems to be the most promising, because it is a prominent technology in
reducing the labour, and providing low production cost, which will be sufficient for
establishing a practical system for in vitro commercialization of mass propagation of
plants.
The term “bioreactor” is generally used to describe a vessel carrying out a biological
reaction, and to refer a reactor vessel for the culture of aerobic cells, or to columns of
packed beds of immobilized cells or enzymes [3]. The bioreactors are widely used for
industrial production of microbial, animal and plant metabolites. The bioreactor
technique applied to plant propagation was first reported by the present author in 1981
on Begonia propagation using a bubble column bioreactor [4]. Since then, bioreactor
83
© 2008 Springer.
www.taq.ir
S. Takayama and M. Akita
technology for plant propagation has developed and aerobic bioreactor culture
techniques have been applied for large-scale production of plant propagules such as
lilies, strawberry, potato, Spathiphyllum, Stevia, etc. [2,5-11]. The bioreactor
technologies are also studied on their characteristics [5,12-16] and on propagation of
several plant species including shoots and somatic embryos [17-29].
The use of bioreactor in micropropagation revealed its commercial applicability, and
recently gained attention to commercial micropropagation process. In this chapter, the
fundamental characteristics in the operation of bioreactor systems and the production of
various plant propagules in bioreactors are described from the standpoint of
bioengineering.
2. Advantages of the use of bioreactor in plant propagation
The use of bioreactor enhances the productivity and the efficiency of plant propagation.
Table 1. Comparison of the specifications of Spathiphyllum propagation between bioreactor
and agar culture.
Items
Bioreactor
Agar culture
Vessel volume
20 L
500 mL
Medium volume L/vessel
16.6 L (liquid)
100 mL (agar)
Number of vessels
6
1000
Number of inocula used for subculture
96 test tubes
150 test tubes
Culture period
90 days
60 days
Equipment
3
Culture space
Number of fluorescent lamps (40W)
0.5 m
6
36 m3
30
Labour
Operational time
200 min
2500 min
*Medium preparation (100 L)
(60 min)
(450 min)
Autoclaving
(10 min)
(140 min)
Inoculation
(45 min)
(1250 min)
Transfer to culture room
(10 min)
(60 min)
Removing cultures
(45 min)
(300 min)
Vessel washing
(30 min)
(300 min)
Transplanting
1800 min
1800 min
*The volume of culture medium was 100 L in both bioreactor culture and agar culture
Such excellent characteristics emerged from the advantages of the use of liquid medium
for plant propagation in the bioreactor and are as follows:
84
www.taq.ir
Bioengineering aspects of bioreactor application in plant propagation
x
Large number of plantlets can easily be produced in one batch in the bioreactor
and scaling up of bioreactor size and number.
x Since handling of cultures such as inoculation or harvest is easy, reducing the
number of culture vessels, and the area of culture space results in the reduction
of costs.
x Whole surface of cultures are always in contact with medium, uptake of
nutrients are stimulated and growth rate is also increased.
x Forced aeration (oxygen supply) is performed which improves the growth rate
and final biomass.
x Cultures are moving in the bioreactor, which results in the disappearance of
apical dominance and stimulates the growth of numerous shoot buds into
plantlets.
In spite of these advantages, there are some pitfalls such as hyperhydricity, plantlet size
variation and microbial contamination [8], etc. The most important problem is the
existence of recalcitrant species for bioreactor application and such species are difficult
to be cultured in liquid medium even if they are possible to be propagated on agar
medium.
These problems need to be rectified and warrants investigation. The efficiency of the
propagation is quite high in the bioreactor compared to solid or shake culture, resulting
in the saving of cost in equipments and labours as indicated in Table 1. After
transplanting in soil, the efficiency of re-establishment of plants during acclimatization
is almost same between the bioreactor and agar cultured plants.
3. Agar culture vs. liquid culture
The plants propagated in a bioreactor are usually submerged in liquid medium. Since
most plants propagated are terrestrial, not aquatic, and under natural habitat, submerged
condition is usually harmful to the plants. In tissue culture, plants can grow under
submerged condition, but this does not mean that plants prefer liquid medium in tissue
culture. The growth response of the plants in liquid medium varied between species or
genera. For example, the growth of Begonia was fairy well in liquid or semi-solid agar
medium (0 to 4 g/L agar) (Figure 1). On the contrary, the growth of Fragaria was
remarkable at solid agar medium (6 to 12 g/L agar), but not in liquid or semi-solid agar
medium. The growth of Saintpaulia revealed the intermediate response between
Fragaria and Begonia (growth was stimulated at 4 to 8 g/L agar). The plants having
hydrophilic nature like Begonia appeared to propagate easily in liquid medium in shake
or bioreactor culture. In spite of the hydrophobic nature, Fragaria plants can grow in
liquid medium in the bioreactor, but require higher aeration rate, and the growth was
linear to aeration rate. In some Clematis species, the growth was strictly repressed in
submerged conditions.
4. Transition from shake culture to bioreactor culture
The shake culture method is considered to be intermediate in establishing bioreactor
techniques. As described above, the growth characteristics in liquid medium are quite
85
www.taq.ir
different between species or genera, so the optimization of culture condition in liquid
medium is the fundamental prerequisite. Once the liquid culture condition is established
in shake culture, the condition can be applied to bioreactor culture for scaling-up.
Figure 1. Effect of agar concentration on growth of Fragaria ananassa(FA), Saintpaulia
ionantha(SI) and Begonia x hiemalis(BH).
5. Types of bioreactors for plant propagation
The bioreactors used for plant propagation are fundamentally the same as that used for
secondary metabolite production by plant, microbial and animal cell cultures. Various
types of bioreactors are used for this purpose which are classified by agitation methods
and vessel construction into; mechanically agitated bioreactors (aeration-agitation
bioreactors, rotating drum bioreactors, spin filter bioreactors), pneumatically agitated
bioreactors (unstirred bubble bioreactor, bubble column bioreactor, air-lift bioreactor),
and non-agitated bioreactors (gaseous phase bioreactor, oxygen permeable membrane
aerator bioreactor, overlay aeration bioreactor) [7]. Mechanically agitated bioreactors
(aeration-agitation bioreactor, Figures 2C, 2D, 2E), the most standardized bioreactor
system in industrial processes, are applicable to plant propagation. However,
pneumatically driven bioreactors such as bubble column (Figure 2A), unstirred bubble
(Figures 2B, 3B) and airlift bioreactors are found to be suitable as plant bioreactors
because it compensate the specific problem of mechanically agitated bioreactors such as
severe shear generation. The most frequently used bioreactors having the characteristics
suitable for plant organs, especially for shoot cultures are unstirred bubble bioreactors,
bubble column bioreactors and airlift bioreactors.
86
www.taq.ir
Figure 2. Various types of bioreactors. (A) Bubble column bioreactor, (B) Unstirred bubble
bioreactor, (C, E) Pilot scale aeration-agitation bioreactor, (D) 10 L aeration-agitation
bioreactor.
6. Preparation of propagules for inoculation to bioreactor
In practical use of bioreactor for plant propagation, large number of propagules which
will be growing to plantlets should be inoculated into the bioreactor. The propagules to
be used as inocula are; multiple shoot buds, regenerative tissues such as protocorm-like
bodies and meristemoids, somatic embryos, and stem or shoot with numerous axillary
buds. Multiple shoot buds which can be obtained by application of cytokinin to the
medium, can be used as propagules for the propagation of most plant species. The
reason of the use of multiple shoot buds is that these cultures are quite stable in their
genetic characteristics. Small pieces of multiple shoot buds cultured in test tubes
containing10 ml agar medium were used as inocula.In case of both Colocasia esculenta
and Spathiphyllum cv. Merry, multiple shoot buds collected from 8 or 16 test tubes
(Figure 3A) were inoculated into 8 L or 16 L unstirred bubble bioreactors, respectively.
After 2 to 3 months of bioreactor culture, a large number of shoots were fully grown in
the bioreactor (Figure 3B). Morphology of inocula and their optimum inoculum size are
87
www.taq.ir
different between genera or species, but usually small inoculum size (1 to 5 g/L) will be
sufficient as inocula in the bioreactor.
Figure 3. Preparation of inoculum in test tubes containing 10 ml of agar medium (A), and
shoot growth in 20 L unstirred bubble bioreactor(B) containing 16 L medium 2 months after
inoculation. The plant is Colocasia esculenta.
7. Characteristics of bioreactor for plant propagation
7.1. FUNDAMENTAL CONFIGURATION OF BIOREACTOR
The bioreactors usually comprise a jacketed pressure vessel which is sterilized by steam
at the beginning of culture and sealed to maintain the sterilecondition during cultivation.
Figure 4 shows the typical aeration-agitation bioreactor vessel generally used for
microbial, animal and plant cell, tissue and organ cultures. The vessel is equipped with
several openings such as an inoculation port, sensor ports (pH, EC, O2, ORP, etc.),
feeding and drain pipes, air inlet and outlet, and so on. These openings should be
completely closed with high quality sanitary fittings and valves. The vessel is also
equipped with heating and cooling jacket which is connected to the steam and water
lines and control the bioreactor temperature. A sealed agitator shaft is inserted in the
bioreactor vessel. The agitator shaft is driven by agitator motor, and impeller(s) is
88
www.taq.ir
attached to the agitator shaft which agitates the culture medium. At the bottom of
bioreactor vessel, air sparger is equipped to circulate the air into the culture medium.
The baffles attached to the vessel wall ensure maximum turbulence during agitation. In
case of shoot propagation, impeller and baffles are detached or mechanical agitation
was stopped to avoid the damage of cultures.
The bioreactor depicted in Figure 4 is quite expensive, which is not realistic for use
in the practical plant propagation. In order to reduce the costs, simplicity of structure
and handling, long-term maintenance of aseptic condition, and of course, sufficient
aeration and mixing are required in design the bioreactor. Practically, a quite simple
bioreactor consist of a vessel with minimum openings using for inoculation, air inlet,
and air outlet, is feasible. Using such a simple bioreactor in batch culture, plants
produced were easily transplanted and established in soil.
Figure 4. Diagram of the structure of typical bioreactor.
89
www.taq.ir
7.2. AERATION AND MEDIUM FLOW CHARACTERISTICS
The characteristics of bubble generation and their hold-up were precisely analyzed by
Aiba et al. [30]. The size of bubbles sparged from orifice of sparger at low aeration rate
was calculated by equation (1);
S
6
3
d B 'Ug
SdV
(1)
Where dB is the diameter of bubbles (mm), d is the diameter of orifice (m), 'U is the
difference of air and liquid density (g/m3), g is the acceleration of gravity (m/s2), and
Vis the surface tension of liquid (dyn/cm) . In equation (1), left-hand side refers to the
buoyancy of bubbles, and the right-hand side is the power equivalent to the retention of
bubbles. This equation was experimentally consistent when aeration rate Q ( cm3 / s)
was within the limits of 0.02 to 0.5 cm3 / sec, and within this limit, the diameter of
bubbles dB (mm) was correlate to d1/3, and not depended on aeration rate Q ( cm3 / s).
Above the limit of Q= 0.5 cm3 / sec, equation (1) was not consistent, and so,
experimental equation (2) was used to estimate dB
dB v Qn
c
(2)
where, n' = 0.2~1.0
A graph on the relationship between diameter of bubbles dB (mm) and superficial gas
velocity VB (m/s) can be split into two parts. When diameter of bubbles was 1.5 mm or
less, the bubbles were mostly spherical, and superficial gas velocity correlated with the
diameter of bubbles. When the ranges of diameter of bubbles were 1.5 to 6 mm, the
bubble shape begins to transform, and superficial gas velocity decrease slightly. When
diameter of bubbles exceeded 6 mm, the bubble shape became mushroom-like
appearance, and superficial velocity correlatively increased with the diameter of bubbles
in the range of 20 to 30 cm/s.
7.2.1. Medium flow characteristics
The medium flow characteristic was influenced by the types of bioreactors. The
direction and velocity of the medium flow severely fluctuated in unstirred bubble
bioreactor (Figure 5A) which reveals the turbulent characteristics and results in the
generation of shear stress. The phenomenon was also remarkable in bubble column
bioreactor. A fundamental solution is the generation of smooth laminar flow of the
medium in the bioreactor. The turbulent characteristics in bubble column or unstirred
bubble bioreactor was changed to smooth laminar flow characteristics when the draft
tube was set in the bioreactor to form airlift bioreactor (Figure 5C). Airlift-like medium
flow was easily attained in unstirred bubble bioreactor by setting the air sparger on one
side at the bottom of the bioreactor (Figure 5B). Although, medium flow near sparger is
turbulent, laminar medium flow is generated partly as shown in Figure 5B.
90
www.taq.ir
The medium flow is characterized by the shape and types of spargers. The straight bar
or ring-shaped brass made sparger with several openings (0.5 to 1 mm diameter)
generate rather large bubbles, and induce turbulent flow nature, but fine bubbles
generated from sintered or ceramic sparger (plate or pipe) induce mild and slow
medium flow. To prevent cell or shoot sedimentation in areas of poor mixing, a plate
shaped sparger made of sintered material at the tapered bottom of bioreactor is effective
[12]. These characteristics indicate the importance of the basic design and construction
of bioreactor in scale-up.
Figure 5. Medium flow characteristics in various types of bioreactors (A) Unstirred bubble
bioreactor, (B) Like (A), but air sparger was set on one side at the bottom of the bioreactor,
(C) Draft-tube airlift bioreactor. AI: air inlet, AO: air outlet, Shadowed region at the
bottom of bioreactor reveal the air sparger.
7.2.2. Medium mixing
The mixing time in relation to shoot fresh weight was measured as shown in Figure 6.
The 10 L unstirred bubble bioreactors containing 8 L medium and Spathiphyllum fresh
shoot grown in the bioreactor, were used for the experiment. Aeration rate was 2 L/min
from a ceramic sparger. Conductometric method using NH4NO3 as salt was used for
determining mixing time.
Medium mixing time without shoot was 18 and 34 s for the unstirred bubble and the
airlift bioreactor, respectively. Increase in mixing time depends on shoot fresh weight
and the type of bioreactor. At the early stage of shoot growth in a bioreactor when shoot
fresh weight was still low (less then 100 g/8L), mixing time was less than 60 s and the
time was shorter in unstirred bubble bioreactor. When shoot fresh weight increased over
100 g/8L, the mixing time delayed exponentially depend on shoot fresh weight
especially in case of unstirred bubble bioreactor. Mixing time became 2190 and 1680 s
91
www.taq.ir
for unstirred bubble and airlift bioreactor, respectively, at highest shoot fresh weight
(2000 g/8L, equivalent to maximum shoot growth in fresh weight).
Figure 6. Relationship between shoot fresh weight in the bioreactor and medium mixing
time. Small graph represent the logarithmic plot in both horizontal and vertical axis.
Aeration: Unstirred bubble bioreactor, Airlift: Draft tube airlift bioreactor.
7.2.3. Oxygen demand and oxygen supply
Plants cultured aerobically require oxygen for growth. In small scale semi-solid
cultures, culture vessels such as flasks or bottles are plugged using gas diffusive
materials. Molecular diffusion through plugs allows oxygen to penetrate into culture
flasks or bottles, and stimulate the cultures to grow. On the contrary, in case of cultures
submerged in liquid medium such as shake or bioreactor culture, natural diffusion of
oxygen is limited and plant growth is strictly inhibited without shaking or forced
aeration. Aeration efficiency evaluated by oxygen transfer coefficient (kLa values)
depends mainly on aeration rate and bubble size [12], and so the type of air sparger is
important to attain higher kLa value. Bubble size generated depends on the type and size
of pores of the sparger. Conventional stainless steel or brass pipe sparger (bar or ring)
with pin holes about 0.5 to 1 mm in diameter is not sufficient for generation of fine
bubbles, and so, to attain sufficient kLa values, aeration rate should be raised. The
requirement of oxygen is different between species and genera, and in general kLa
values over 10 h-1 is sufficient for growth in cultures of many plant species. For
example, in case of tobacco cell cultures, the final biomass concentration became
constant at kLa values over 10 h-1 [31]. But when KLa was set under 10 h-1, cell yield
became depended on KLa values [31]. The factors which affect KL and a are the mixing
conditions in the bulk liquid, the diffusion coefficient, the viscosity and the surface
tension of the medium, air-flow rate, gas hold-up and the bubble size [32]. The specific
interfacial mass transfer coefficient KL is constant for fixed medium and temperature
and is relatively insensitive to the fluid dynamics in the bioreactor [33], but the specific
interfacial area a is difficult to measure, and so the two parameters are combined and
referred to as the volumetric mass transfer coefficient, KLa. The difference in KLa is
mainly attributed to differences in the specific interfacial area a which was affected by
92
www.taq.ir
aeration rate, size of bubbles, and mixing. KLa values are also affected by types of
bioreactor and diameter of draft tube. In the scale up of airlift bioreactor, the long
residence time of small air bubbles in tall columns may lead to the depletion of oxygen
from these bubbles which resulted in the decline of KLa [34]. A need of higher KLa
values was also evident in the shoot culture of strawberry in a bioreactor, where the
growth of shoots correlated to kLa values and to aeration rate [35]. A problem in higher
aeration is the generation of higher mechanical stress by turbulent agitation (shear
stress). In order to enhance the aeration efficiency without the generation of severe
shear stress, the use of ceramic or sintering steel porous sparger is effective, which
generate the fine bubbles with higher KLa values (Figure 7).
Figure 7. Effect of the types of air sparger and aeration rate on oxygen transfer coefficient
in unstirred bubble bioreactor containing 6 L liquid medium. Oxygen transfer coefficient
was expressed as KLa (h-1).
7.3. LIGHT ILLUMINATION AND TRANSMITTANCE
Production of plants with well developed and green leaves in the bioreactor is preferable
for re-establishment of the plants in soil. The production of such plants depends mainly
on the intensity of illumination to the cultures. Illumination of propagules in a bioreactor
is not easy because of the logarithmic reduction of light intensity passing through the
plant tissues and the distance from light source. Figure 8 indicates the relationship
between the distance of a source of light and its intensity. Light emitting diode (LED) is
superior to other light sources because of its excellent focusing characteristics, i.e. the
high energy conversion rate and reduced infrared heat radiation. Light transmittance was
reduced drastically by the presence of shoot cultures in the bioreactor especially at
higher fresh weight (Figure 9). When shoot cultures of Spathiphyllum and Colocasia
grown in a bioreactor made of glass vessel were illuminated externally by fluorescent
lamps, light transmitted to the cultures only several centimetres from the vessel surface.
The leaves on illuminated shoots became green and well developed. On the other hand,
93
www.taq.ir
the cultures growing in the bioreactor were etiolated and leaf expansion was inhibited.
The same phenomenon was also observed in shoot cultures of Stevia grown in large
scale (500 L) bioreactor equipped with 4 lamps [36,37]. Although various illuminated
bioreactors have been designed [38,39], application to commercial propagation is
limited because the price becomes expensive and light introduction was not efficient.
Development of new culture technology for propagation in the bioreactor with high
illumination efficiency, or production of transplantable propagules in the bioreactor
without or with low illumination is required.
Figure 8. Relationship between the distance of a source of light and its intensity. I0: light
intensity at the surface of light source. I: light intensity measured at certain distance.
Figure 9. Relationship between light path length and light transmittance in various degree
of shoot growth in cultures of Spathiphyllum. I0: light intensity at the surface of shoot
cultures I: light intensity measured at certain path length.
94
www.taq.ir
8. Examples of bioreactor application in plant propagation
Many plant species and varieties have been cultured in the bioreactor [2,5,7,8,40].
Responses of cultures in bioreactors are quite different among species or genera and
they could be also different from the responses observed under static culture conditions
on semi-solid medium (see section 7.2). The cultures propagated were regenerated from
inoculated cultures consists of multiple shoot buds induced by the addition of cytokinin
to the medium. During cultivation in the bioreactor, various types of plant propagules
such as shoots, bulbs, microtubers, corms, embryos, etc. are possible to be developed
from shoot buds. The propagules produced in the bioreactor should be easily adapted to
ex vitro conditions as possible. Storage organs such as bulbs, corms or tubers seem to
be the best choice for proliferation in bioreactors. Several examples of bioreactor
applications for plant propagation are listed as follows:
x Shoots: Atropa belldona, Begonia x hiemalis, Chrysanthemum morifolium,
Dianthus caryophyllus, Fragaria ananassa, Nicotiana tabacum, Petunia
hybrida, Primula obconica, Zoysia japonica, Scopolia japonica,
Spathiphyllum, Stevia rebaudiana, etc.
x Bulbs: Fritillaria tunbergii, Hippeastrum hybridum, Hyacinthus orientalis,
Lilium, etc.
x Corms: Caladium sp., Colocasia esculenta, Pinellia ternate, etc.
x Tubers: Solanum tuberosum
x Embryos or adventitious buds: Atropa belladona
9. Aseptic condition and control of microbial contamination
The microbial contamination is frequently observed in laboratory and commercial tissue
cultures, and sometimes leads to the severe damages to cultures. The cause of microbial
contamination is latently expressed pathogenic or plant-associated micro-organisms and
laboratory contaminants associated with the operatives and in both cases, microbes are
expressed in any culture stage [41]. The microbial contamination observed in the
laboratory processes is influenced by various factors but the problem is overcome by
aseptic handling of vessels, equipments, and cleanliness of culture room, as well as the
skilfulness of the operators. The extensive problem of microbial contamination is
caused by the proliferation of mites. The mites quickly proliferate and spread around
and invade the culture vessels [8]. The seed cultures of propagules used as inocula are
sometimes invaded by mites, and cause the contamination after inoculation into
bioreactor. To avoid these problems, periodical fumigation of culture room should be
performed, and it is strongly recommended that stock cultures are maintained in test
tubes with spongy silicon plugs [8].
Several factors intimately relating to microbial contamination are conceivable
[42,43] especially hardware design, construction and manipulation manner. To avoid
contamination, bioreactor construction should be made simple. The number of tube
connectors and various openings of the culture vessel such as the inoculation port
should be minimized. In addition, pre-sterilization of empty bioreactor vessel at 121oC
for 30 min is usually necessary. Then bioreactor filled with the culture media should be
sterilized again at 121oC for 15 min. The inoculation is the risky process because
95
www.taq.ir
bioreactors always exposed to external air conditions. The inoculation of the seed
culture of propagules to portable sized bioreactors is performed in laminar flow clean
air bench. In an open air condition, especially when the bioreactor is anchored to the
floor, inoculation should be done in burning flames of alcohol or gas-burner completely
covering the inoculation port. In case of large-scale bioreactor (500 L) which is
anchored to the floor, Kawamura et al. [44] developed an apparatus to inoculate a large
number of plantlets or tissue segments. The use of such equipment results in reduction
of microbial contamination.
Aeration is also the cause of microbial contamination. Autoclavable heat-resistant
tubes and disposable ultra-filter (pore size; 0.2 to 0.45 µm) are adopted as materials in
the air line. An air outlet is sometimes equipped with glass wool filter which was wetted
by the splash of culture medium and cause the invasion of aphids and microbes. A
simple solution is the use of spiral tube (about one meter) with cut end, which prevents
the invasion of microbes.
10. Scale-up to large bioreactor
10.1. PROPAGATION OF STEVIA SHOOTS IN 500 L BIOREACTOR
The advantage of the use of bioreactor for plant propagation is the easiness in scale-up.
The example is the use of 500 L bioreactor for Stevia rebaudiana propagation (Figure
10,11) [35,36]. The cluster of shoot primordia which were propagated in the shake
culture using modified MS medium (half-strength of KNO3, NO4NO3 and CaCl2,2H2O
were used), supplemented with 0.1 mg/L NAA, 1 mg/L BA and 30 g/L sucrose, was
used as inocula. The 500 L bioreactor contained 300 L MS medium supplemented with
10 g/L sucrose, sterilized at 120oC for 30 minutes by direct application of steam at 0.1
MPa. The fresh weight of shoot buds as inocula was 460 g. Cultures in a bioreactor was
aerated at 15 L/minutes, illuminated at 16 h photoperiod by 4 fluorescent lamps inserted
in the bioreactor, and incubated at 25oC for 1 month. During the culture period, at 3
weeks, 20 L of the medium was removed and newly prepared 50 L of the same medium
containing 6,300 g sucrose was added to elevate the consumed nutrient, sucrose and
water. The shoots grew actively to fill up culture vessels within one month. The total
shoot weight was 64.6 kg in fresh weight, which was 140 times the inoculum weight.
The growth efficiency in 500 L bioreactor was almost the same as in shake culture (100
ml medium in 300 ml flask) or in 10 L bioreactor (6 L medium in 10 L bioreactor). The
shoots adjacent to fluorescent lamps were green and developed leaves, but most shoots
were etiolated and leaf development was significantly suppressed because the light
intensity exponentially decreased with distance under high plantlet density (Figure 10).
The shoots taken out from the bioreactor had no roots, but could be easily acclimatized,
and after transplant in soil, more than 90% of number of the shoots was successfully
acclimatized in soil. These results indicate the practical applicability of large scale
propagation using bioreactor.
96
www.taq.ir
Figure 10. Propagated shoots of Stevia taken out from the bioreactor (a, green shoots
growing aound the fluorescent tube; d, completely white shoots growing remote from
fluorescent tubes; b and c, intermediate location of a and d.
Other types of large bioreactors were also applicable. For example, Stevia rebaudiana
shoots were propagated using a separated impeller-type 500 L bioreactor (Figure 11).
Shoots were also well grown in this type of bioreactor and harvest of unwounded
cultures was much easier than the case described above.
Figure 11. Large scale propagation of Stevia rebaudiana shoot in a separated impeller-type
500 L bioreactor. (A) Diagram of a separated impeller-type bioreactor used for mass
propagation of Stevia rebaudiana. Shoots were cultured in a bioreactor illuminated with
fluorescent lamps. Fluorescent tubes equipped within the bioreactor were abbreviated in
this figure. (B) Shoot cultures in a bioreactor illuminated with fluorescent lamps, (C) Whole
view of Stevia rebaudiana shoot cultures adhered to cylindrical mesh which was taken out
from bioreactor.
97
www.taq.ir
10.2. SAFE INOCULATION OF PLANT ORGANS INTO BIOREACTOR
As described previously, the most risky process to microbial contamination is the
inoculation of seed cultures. In general, microbial or plant cell suspension as seed (seed
culture) is previously cultured in a smaller size bioreactor and transferred through
inoculation tube or pipe connecting between bioreactors during an inoculation.
Application of this simple method is difficult in case of plant propagation in the
bioreactor because of blockage of the tube by inoculated tissue segments. The tissue
segments frequently used as inocula for production of propagules are shoots,
adventitious buds, axillary buds, bulbscales. These tissue segments are usually
inoculated through inoculation port. The bioreactors of 1 to 20 L are settled in clean
bench, and inocula are transferred into bioreactor through inoculation port using
forceps. It is better to cover the inoculation port in flames using methyl alcohol or ring
burner. In case of large-scale bioreactor anchored on the floor of pilot plant, use of a
sanitary apparatus for inoculating a large number of plant propagules is promising.
11. Prospects
The use of liquid systems especially the bioreactor technique seems to be successfully
applicable in commercial propagation, and actually a part of tissue culture nurseries
already adopted this technique. However, at present, many problems still exists for wide
application of this technique. The growth conditions in bioreactor are somewhat
different from agar culture and it is necessary to find the optimum culture condition in
the liquid medium. Skill is also required in handling and operating the bioreactors as
well as in preparation of large number of aseptic seed cultures in one batch. Although it
is possible to produce several types of organs in bioreactors, propagation of storage
organs will be the best choice for proliferation, because the culture process is quite
simple, and the produced propagules are easy to handle and suitable for acclimatization.
The bioreactor technology is advantageous in their proven high efficiency and easiness
of operation process, and appears to be the most promising system for industrial plant
propagation.
References
[1] Altman, A. and Loberant, B. (2000) Micropropagation of plants, principles and practices. In: Spier, R.E.;
Griffiths, B. and Scragg, A.H. (Eds.) The Encyclopaedia of Cell Technology. ISBN: 0-471-16123-3,
John Wiley & Sons, Inc., New York; pp.916-929.
[2] Takayama, S. (1991) Mass propagation of plants through shake and bioreactor culture techniques. In:
Bajaj, Y.P.S. (Ed.) Biotechnology in Agriculture and Forestry. Vol.17. Springer-Verlag, Berlin; pp. 495515.
[3] Coombs, J. (1986) MacMillan Dictionary of Biotechnology. Macmillan Press, London; pp. 1-330.
[4] Takayama, S. and Misawa, M. (1981) Mass propagation of Begonia hiemalis plantlets by shake culture.
Plant Cell Physiol. 22: 461-468.
[5] Takayama, S. (2002) Practical aspects of bioreactor application in mass propagation of plants. Abst. 1st
Int. Symp. Liquid Systems for In Vitro Mass Propagation of Plants. Norway, May 29th – June 2nd, 2002.
pp. 60-62.
98
www.taq.ir
[6] Takayama, S.; Arima, Y. and Akita, M. (1986) Mass propagation of plants by fermentor culture
techniques. In: Abst. 6th International Congress of Plant Tissue and Cell Culture, Int. Assoc. Plant Tissue
Cult. University of Minnesota. p. 449.
[7] Takayama, S. and Akita, M. (1994) The types of bioreactors used for shoots and embryos. Plant Cell
Tissue Org. Cult. 39:147-156.
[8] Takayama, S. and Akita, M. (1998) Bioreactor techniques for large-scale culture of plant propagules. Adv.
Hort. Sci. 12: 93-100.
[9] Akita, M. and Takayama, S. (1994) Induction and development of potato tubers in a jar fermentor. Plant
Cell Tissue Org. Cult. 36: 177-182.
[10] Akita, M. and Takayama, S. (1994) Stimulation of potato (Solanum tuberosum L.) tuberization by semicontinuous liquid medium surface level control. Plant Cell Rep. 13: 184-187.
[11] Akita, M. (2000) Bioreactor culture of plant organs. In: Spier,R.E.; Griffiths, B. and Scragg, A.H. (Eds.)
The Encyclopaedia of Cell Technology. ISBN: 0-471-16123-3, John Wiley & Sons, Inc., New York;
pp.129-138.
[12] Takayama, S. (2000) Bioreactors, Airlift. In: Spier, R. E.; Griffiths, B. and Scragg, A.H. (Eds.) The
Encyclopaedia of Cell Technology, ISBN: 0-471-16123-3, John Wiley & Sons, Inc., New York; pp. 201218.
[13] Archambault, J.; Williams, R.D.; Lavoie, L.; Pepin, M.F. and Chavarie, C. (1994) Production of somatic
embryos in a helical ribbon impeller bioreactor. Biotechnol. Bioeng. 44: 930-943.
[14] Archambault, J.; Lavoie, L.; Williams, R.D. and Chavarie, C. (1995) Nutritional aspects of Daucus
carota somatic embryo cultures performed in bioreactors, In: Terzi, M.; Cella, R. and Falavigna, A.
(Eds.) Current Issues in Plant Molecular and Cellular Biology. Kluwer Academic Pulblishers, Dordrecht,
[15] Heyerdahl, P.H.; Olsen, O.A.S. and Hvoslef-Eide, A. K. (1995) Engineering aspects of plant propagation
in bioreactors. In: Aitken-Christie, J.; Kozai, T. and Smith, L.M. (Eds.) Automation and Environmental
Control in Plant Tissue Culture. Kluwer Academic Publishers, Dordrecht, The Netherlands; pp.87-123.
[16] Ziv, M. (2000) Bioreactor technology for plant micropropagation. Hort. Rev. 24:1-30.
[17] Ammirato, P.V. and Styer, D.J. (1985) Strategies for large scale manipulation of somatic embryo in
suspension culture, In: Zaitlin, M.; Day, P. and Hollaender, A. (Eds.) Biotechnology in Plant Science:
Relevance to Agriculture in Eighties. Academic Press, NewYork; pp. 161-178.
[18] Harrell, R.C.; Bieniek, M.; Hood, C.F.; Munilla, A.R. and Cantliffe, D.J. (1994). Automated in vitro
harvest of somatic embryos. Plant Cell Tissue Org. Cult. 39:171-183.
[19] Jay, V.; Genestier, S. and Courduroux, J.C. 1994. Bioreactor studies of the effect of medium pH on
carrot (Daucus carota L.) somatic embryogenesis. Plant Cell Tissue Org. Cult. 36:205-209.
[20] Levin, R.; Gaba, V.; Tal, B.; Hirsch, S.; Denola, D. and Vasil, I.K. (1988) Automated plant tissue culture
for mass propagation. Bio/Technol. 6: 1035-1040.
[21] Preil, W.; Florek, P.; Wix, U. and Beck, A. (1988) Towards mass propagation by use of bioreactors.
Acta Hort. 226: 99-105.
[22] Preil, W. (1991) Application of bioreactors in plant propagation. In: Debergh, P.C.; Zimmerman, R.H.
(Eds.) Micropropagation Technology and Application. VIII, Kluwer Academic Publishers Group,
Boston, USA; pp. 425-446.
[23] Styer, D.J (1985) Bioreactor technology for plant propagation. In: Henke, R.R.; Gher, K.W.;
Constantin,J. and Hollander. A. (Eds.) Tissue Culture in Forestry and Agriculture. Plenum Press, New
York; pp.117-130.
[24] Tautorus, T.E.; Lulsdorf, M. M.; Kikcio, S.I.. and Dunstan, D.I. (1994) Nutrient utilization during
bioreactor culture and maturation of somatic embryo cultures of Picea mariana and Picea glaucaengelmannii. In Vitro Cell. Dev. Biol.- Plant 30: 58-63.
[25] Wheat, D.; Bondaryk, R.P. and Nystrom, J. (1986) Spin filter bioreactor technology as applied to
industrial plant propagation. Hort. Sci. 21:819.
[26] Ziv, M. (1990) Morphogenesis of gladiolus buds in bioreactors - Implication for scaled-up propagation
of geophytes. In: Nijkamp, H.J.J.; Van Der Plas, L.H.W.; Artrijk, J. V. (Eds.) Progress in Plant Cellular
and Molecular Biology. Kluwer Academic Publishers, Dordrecht, The Netherlands; pp. 119-124.
[27] Ziv, M. (1995) The control of bioreactor environment for plant propagation in liquid culture. Acta Hort.
393: 25-38.
[28] Ziv, M. and Shemesh, D. (1996) Propagation and tuberization of potato bud clusters from bioreactor
culture. In Vitro Cell. Dev. Biol.- Plant 32: 31-36.
[29] Ziv M.; Kahany, S. and Lilien-Kipnis, H. (1994) Scaled-up proliferation and regeneration of Nerine in
liquid cultures: Part I. The induction and maintenance of proliferating meristematic clusters by
paclobutrazol in bioreactors. Plant Cell Tissue Org.Cult. 39: 109-115.
.
99
www.taq.ir
[30] Aiba, S.; Humphrey, A.E. and Millis, N.F. (1965) Biochemical Engineering. University of Tokyo Press;
pp. 1-345.
[31] Kato, A.; Shimizu, Y. and Noguchi, S. (1975) Effect of initial KLa on the growth of tobacco cells in
batch culture. J. Ferment. Technol. 53: 744-751.
[32] Fonseca, M.M.R.; Mavituna, F. and Brodelius, P. (1988) Engineering aspects of plant cell culture. In:
Pais, M.S.S.; Mavituna, F. and Novais, J.M. (Eds.) Plant Cell Biotechnology. Springer-Verlag, Berlin;
pp. 389-401.
[33] Blenke, H. (1979) Loop reactors. Adv. Biochem. Eng. 13: 121.
[34] Payne, G.F.; Shuler, M.L. and Brodelius, P. (1987) Large scale plant cell culture. In: Lydersen, B.J. (Ed.)
Large Scale Cell Culture Technology. Carl Hanser Verlag, Munich. ISBN 3-446-14845-0; pp. 193-229.
[35] Takayama, S.; Amo, T.; Fukano, M., and Oosawa, K. (1985) Mass propagation of strawberries by jar
fermentor culture. (2) Studies on the optimum conditions in a liquid medium and the establishment of
mass propagation scheme using a jar fermentor. Abst. 1985 Spring Meeting of J. Soc. Hort. Sci. Tokyo;
pp. 210-221.
[36] Akita, M.; Shigeoka, T.; Koizumi, Y. and Kawamura, M. (1994) Mass propagation of shoots of Stevia
rebaudiana using a large scale bioreactor. Plant Cell Rep. 13: 180-183.
[37] Akita, M.; Shigeoka, T.; Koizumi, Y. and Kawamura, M. (1994) Mass propagation of multiple shoots
using a large bioreactor. J. Soc. High Technol. Agric. 6: 113-121.
[38] Ikeda, H. (1985) Culture vessel for photoautotrophic culture, Japan Patent, Kokai. 60-237984.
[39] Inoue, H. (1984) Culture vessel for photo-requiring organisms, Japan Patent, Kokai. 59-21682.
[40] Takayama S.; Arima, Y. and Akita, M. (1986) Mass propagation of plants by fermentor culture
techniques. Abst. 6th Int. Cong. Plant Tissue Cell Cult., Int. Assoc. Plant Tissue Cult., University of
Minnesota, Minnesota, USA; p. 449.
[41] Cassels, A.C. (1991) Control of contamination in automated plant propagation. In: Vasil, I. K. (Ed.) Cell
Culture and Somatic Cell Genetics of Plants. Academic Press, New York. ISBN.0-12-715008-0. 8:197212.
[42] Manfredini, R.; Saporiti, L.G. and Cavallera, V. (1982) Technological approach to industrial
fermentation: limiting factors and practical solutions. La Chimica e Industria. 64: 325-334.
[43] Takayama, S. (1997) Bioreactors for plant cell tissue and organ cultures, In: Vogel, H.C. and Todaro,
C.L.(Eds.), Fermentation and Biochemical Engineering Handbook. 2nd Edition, Noyes Publications,
Westwood, New Jersey, USA; pp. 46-70.
[44] Kawamura, M.; Shigeoka, T.; Akita, M. and Kobayshi, Y. (1996) Newly developed apparatus for
inoculating plant organs into large-scale fermentor. J. Ferment. Bioeng. 82: 618-619.
100
www.taq.ir
AGITATED, THIN-FILMS OF LIQUID MEDIA FOR EFFICIENT
MICROPROPAGATION
JEFFREY ADELBERG
Department of Horticulture, Clemson University, Clemson SC, USA,
29634 - Fax: 864-656-4960 - Email: [email protected]
1. Introduction
In vitro culture is a semi-closed system that aseptically provides oxygen, water, organic
carbon source (and/or CO2 and light), nutrients, and plant growth regulators (PGR), at a
controlled temperature. A traditional view of plant tissue culture involves placing a
small piece of tissue on the gelled-media surface, in a jar, plate or tube, and allows
exponential growth unfettered by lack of resource in a uniform microenvironment.
Many reports summarized in this volume show increased productivity (per plant, unit
area or time) were achieved with larger vessels of liquid medium yielding greater
numbers and / or larger plants. Liquid systems that improve distribution of dissolved
nutrients, water and oxygen, in the vessel stimulate growth of plant tissues. Simplicity,
cost and ergonomic factors are human constraints imposed on designs intended for
commercial use.
This chapter describes a hybrid micropropagation process that invokes features of
semi-solid gel and bioreactor technology. The agitated, thin-film system (or rocker)
uses large, rigid rectangular vessels in a slow pitching motion to intermittently wet and
aerate plantlets [1] (see Figure 1). Economy of scale was optimized for the twodimensional growth surface area in the vessel. Gentle oxygenation of liquid media was
similar to wave machines Eibl and Eibl describe for cell and tissue culture in Part 2 of
this volume. Shoot surfaces, intermittently wet or dry in a large headspace, accumulate
large quantities of solutes from media resulting in high shoot quality similar to
temporary immersion systems. Vessel and culture room designs differ from
conventional micropropagation, or the bioreactors discussed in other chapters of Part 2.
The first section of this chapter discusses nutrients and heterotrophic growth in agar and
liquid;the secondsectioncomparesefficiencyof agitated,thin-film processwith agar-based
media system; and third one lists designconsiderations for the vessel and culture shelves
in the growth room during scale-up. Comparisons will be drawn to agar-gelled media in
small round jars, typical of many micropropagation protocols using semi-solid media.
101
© 2008 Springer.
www.taq.ir
J. Adelberg
Figure 1. Agitated thin films are created by slowly pitching large rectangular vessels.
Reproduced from Adelberg, J. (2004) [24] with permission from Society for In Vitro
Biology.
2. Heterotrophic growth and nutrient use
2.1. SOLUTES IN SEMI-SOLID AGAR
Heterotrophic plant growth depends on the uptake of sugar, water, and nutrients from
medium. Agar, or other organic gelling agents, are frequently used despite problems of
mineral impurities, limited hydraulic conductance, limited availability of solutes to the
tissue and binding of toxic exudates near the tissue interface [2,3,4]. Solute movement
through gelled media and transfer to the plant is primarily by diffusion [5]. Uptake at
the interface surface may proceed against concentration gradients at latter stages of the
culture cycle when active uptake by roots and callus is likely to occur. The sealed
culture vessel with high humidity limits transpiration, restricting mass flow of dissolved
solutes through the xylem and intercellular space.
Selecting an optimal plant density is of great importance to system efficiency but
creates a trade-off between productivity and plant quality. Greater plant densities per
volume of medium increased the uptake of macro-nutrients, including sucrose, for four
ornamental perennial crops; Delphinium, Iris, Hemerocallis, and Photinia [6,7]. Highest
102
www.taq.ir
Agitated, thin films of liquid media for efficient micropropagation
plant densities had the lowest multiplication rates and the lowest rate of nutrient uptake
per plant. However, the greatest yield of new plants per vessel per unit time was derived
at high plant densities. Nutrient availability in high-density agar-gelled cultures was a
limitation to multiplication. Nitrate, phosphate and sugar uptake of single plantlets in
test tubes of static liquid media greatly exceeded what would be available to plantlets in
a normal density for commercial propagation on agar with Hemerocallis and
Delphinium [5].
Sucrose is the solute supplied in the largest quantity in most tissue culture media,
having both nutritive and osmotic effects on plant growth. Ibaraki and Kurata [8]
described the movement of sucrose in their adjacent medium model as a series of three
resistance components: a) diffusion across the medium following a Fick's law equation
with the diffusion coefficient specific to solute/solvent, b) boundary layer resistance at
the interface surface of the plant and medium, and c) resistance in the plant tissue
corresponding to the biochemical sink strength and the plant's transport properties.
Diffusion in medium requires calculating the one-dimensional concentration gradients
in sugar concentration with time. Sucrose moves approximately 4-times faster in
stationary water than agar gel. The boundary resistance at the plant/medium interface
was approximately 6000 times greater in agar than liquid media per unit surface area. It
is easily envisioned that a plantlet impinged on the surface of an agar gel has a much
smaller surface area for exchange at its base than a similar plantlet wet with nutrient
across its entire surface. Ibaraki and Kurata [9] further developed a heterotrophic
growth model that simulated fresh and dry weight based on water and sugar uptake. Dry
matter accumulation was determined by the difference between sugar levels in medium
and plant at the interface surface and the area of that surface. Fresh weight gain is
related to the plants relative water content, the water content of the medium, and the
interface surface area.
Positional non-equilibrium of sugar concentration residual in vessels of spent media
(Table 1) suggests uptake by the plant may exceed replenishment across the gel. There
was significantly less sugar adjacent to the plantlet compared to media in a distal
position. Species and genotypes had different quantities of sugar uptake relative to sink
strength and the plants' internal transport properties. Benzyladenine concentrations
affected the rate of sugar uptake differently among the genotypes. Hypothetically,
increasing the size of the vessel, the duration of the culture cycle, or the density of
plants in the gelled media would increase the magnitude of the non-equilibrium. It is
also likely that compounds less soluble than sugar would experience greater nonequilibrium at the conclusion of the culture cycle. There is a lack of experimental data
published on the diffusion of common ions in agar media.
2.2. SOLUTES IN STATIONARY LIQUIDS
Stationary liquid culture (e.g. floated tissue, membrane rafts, paper bridges, foam cubes)
are useful for research scale nutrient experiments, but not generally useful for large-scale
propagation. Interface surface areas are roughly equivalent to what may be found with
gelled media and liquid is suited to repeated sampling to develop time course data. In
one such experiment, when floated leaf disks of tobacco were assayed for uptake of
eight nutrient ions during 5-weeks on hormone free media, only iron uptake was
significant. When shoot organogenesis was stimulated by benzyladenine, nitrate,
103
www.taq.ir
J. Adelberg
phosphorous, potassium and sulphur uptake became significant following a 10-day lag
phase, associated with meristem initiation and shoot growth [10]. Nitrate and
phosphorous residual concentrations in media approached zero near termination of the
culture cycle.
Table 1. Sugar used from MS media containing two concentrations of benzyladenine, 30 g/l
sucrose 0.7% agar solidified media after 5-weeks of culture. Over 300, 180-ml baby food
jars containing gelled media were assayed at positions distal and adjacent to the base of the
growing plantlet.
Sugar used (g/l)
1 µM BA
5 µM BA
Genotype and species
Distala
Adjacentb.
Distal
Adjacent
Hosta 'Blue Mammoth
4.6 ± 0.9
8.1 ± 0.9
4.2 ± .0.7
5.6 ± 0.4
Hosta 'Francee'
8.5 ± 1.2
11.0 ± 0.7
4.9 ± 0.8
7.6 ± 0.7
Hosta 'Great Expectations'
6.5 ± 1.0
8.9 ± 1.2
3.3 ± 1.7
2.4 ± 1.6
Hosta 'Hadspen Blue'
0.2 ± 0.9
1.1 ± 0.9
7.4 ± 1.0
7.6 ± 1.3
Hosta 'Shade Fanfare'
1.1 ± 1.1
-0.3 ± .6
2.0 ± 1.0
1.9 ± 1.3
Hosta 'Inniswood'
10.5 ± 0.8
10.6 ± 1.3
7.5 ± 1.6
9 ± 1.5
Hosta 'Wide Brim'
15.6 ± 1.3
15.8 ± 1.2
16.9 ± 1.2
17.3 ±1.5
Colocasia antiquorum 'Illustris'
8.0 ± 3.2
12.0 ± 2.6
13.8 ± 1.8
14.2 ± 1.2
Zingiber miyoga 'Danicing Crane'
-5.0 ± 0.2
-4.7 ± 0.3
-5.0 ± 0
1.9 ± 1.3
a. Media was sampled at harvest time with a pipette on the outer perimeter surface of media in the vessel,
approximately 1 cm from the nearest plant's base.
b. Media was sampled at harvest time with pipette directly underneath the harvested plants.
Time-course studies of solutes use was conducted on membrane rafts that created a
liquid interface surface area similar to that found in agar-based culture. Axillary bud
proliferation of watermelon with high concentrations of benzyladenine resulted in
ammonium depletion related to cessation of growth over 5-week period [11]. Lowered
benzyladenine concentrations and added gibberellic acid caused shoot elongation with
increased growth. Ammonium depletion was associated with cessation of growth and
there was an increased uptake of nitrate, calcium and potassium, related to greater fresh
weight. There was an inverse correlation between plant biomass, and residual
concentrations of sugar, ammonium, nitrate, potassium, calcium, and direct correlation
of biomass to water use (Table 2). Refractive index, measured in BRIX, is a rapid,
inexpensive measurement with no expendable reagents and real time feedback. Decline
in BRIX may be used to monitor plant growth or nutrient ion uptake in repeated batch
processes. Plant cells have roughly 50% conversion efficiency of organic carbon feed to
final cell dry weight [12]. Patterns of specific nutrient ion use may change dependent on
developmental stage and under the influence of plant growth regulators.
104
www.taq.ir
Table 2. Correlation coefficients of biomass (fresh and dry weight) with nutrient depletion
and water use of watermelon shoot cultures in elongation medium on polypropylene
membrane rafts at six sampling dates during 38-day time course experiment.
BRIXa
Waterb
Ca+2 c
K+ c
N03- c
NH4+ c
Dry weight
-0.98
0.91
-0.84
-0.93
-0.98
-0.89
Fresh weight
-0.98
0.90
-0.81
-0.93
-0.98
-0.88
Dry/fresh
0.63
-0.55
0.52
0.60
0.65
0.63
a. Residual sugar in media measured with refractometer.
b. Volume of water used from media determined by volume of residual medium, adjusted for water loss from
vessel by evaporation.
c. Concentration of ion in residual medium determined by ion-selective electrode by method described by
Desamero et al. (1993).
Primarily, fresh weight gain during heterotrophic plant culture is due to the uptake of
water and the dry weight gain is due mainly to the uptake of sugar and inorganic ions.
Plants from agar and stationary liquid cultures had similar fresh and dry weights for
Venus flytrap (Drosera muscipula). Relative dry matter of plants (dry weight / fresh
weight) was inversely correlated to concentration of sugar in residual media at time of
harvest (Figure 2). Plants grown at higher densities (5x difference from high to low) had
lower residual sugar concentrations, on both agar and liquid. Also, cultures with more
sucrose (5% vs. 3% w/v) used more sucrose, but had greater residual sugar
concentrations. In both agar and liquid with 3% sucrose, relative dry matter was reduced
from 11.5% to 9.3% by increased plant density, and in 5% sucrose medium relative dry
matter was reduced from 19.6% to 13.8% in response to increased density. Water
uptake depends upon the water potential difference between the plantlet and medium
[9]. When sugar becomes depleted at high densities, plants continue to grow by taking
on more water relative to soluble solids. Increased sugar concentrations allow higher
density cultures to maintain high relative dry matter content.
2.3. SUGAR IN SHAKER FLASKS AND BIOREACTORS
In shake-flask culture, the entire plant surface is available for nutrient exchange.
Turbulent media does not develop gradients and there is less resistance to solute
transfer. Oxygenation of media by shaking creates shear forces that damage many plant
tissues, but a few species are suited to research scale micropropagation experiments.
When Cymbidium protocorm-like bodies (PLB's) were micropropagated in shake-flask
culture withglucose concentrations in medium ranging from 0.1 -2% (w/v) fresh and dry
weight increased with sugar concentration. The rate of dry weight accumulation per unit
surface area remained relatively constant with PLB's having 7- 14% relative dry weight.
However,the fresh weight gain per unit surface area was inversely related to relative dry
weight because plants with high relative dry weights have a greater influx of water [9].
105
www.taq.ir
J. Adelberg
Figure 2. Correlation between residual sugar in media and relative dry weight of Venus
flytrap, Drosera muscipula following five weeks in stationary culture under varied
conditions. Vessels were initiated for agar and liquid medium, with 3 and 5% w/v sucrose
over a range of explant densities. Each data point represents tissue sampled from one
vessel.
With Hosta plantlets in shake-flasks, initial levels of sucrose in media from 1-7% w/v
were directly related to endogenous levels of sucrose, glucose and fructose following 5weeks of culture [13]. Shoot bud multiplication was optimal at 5% media sucrose. As
sucrose was increased from 1-7% (w/v), shoot and root dry weights increased linearly in
roots as did shoots in medium containing benzyladenine, but in hormone-free medium
dry weight gain levelled at 5% sucrose (w/v). Media sucrose at stage II was related to
greater dry weight, lowered mortality and less leaf chlorosis, following rooting, coldstorage for 7 or 14 weeks, and re-growth in greenhouse [14]. Modelling sugar uptake,
translocation, storage and re-growth could be developed to maximize values of young
plants for shipping and in international commerce.
Specialized storage organs of geophytes, bulbs, corms, tubers, rhizomes, are
modified shoot systems with reduced stem and leaf surfaces. Bioreactor and shaker
systems are well suited for large-scale micropropagation of micro-scaled storage organ
in many geophytes including lily [15], garlic [16], potato [17], turmeric [18], and taro
[19]. Liquid medium with high sugar concentrations (5-12% w/v) results in higher dry
weights and stored carbohydrate related to better quality planting stock. Heterotrophic
growth models of storage organ culture would assist in assigning value to products of
bioreactor process.
With leafy shoot systems, temporary immersion (TIS), or partial immersion, with
correctly timed cycles avoided hyperhydricity, limited shear force, provided adequate
oxygen, and sufficient mixing of medium. Larger plants, with superior shoot quality in
TIS, comparedto agar are documented for many unrelated species [20] (see also Afreen,
F. in this book). In one such comparison of pineapple shoots from conventional agar
106
www.taq.ir
and TIS, TIS shoots were larger with greater leaf area with more dry weight, due to an
approximately 10-fold increase in sugar and nitrate assimilation on a fresh weight basis
[21].
Table 3. Multiplication rate and number of new plants per square meter of bench space per
week generated in agar containing baby food jars and large, rectangular vessels in
agitated, thin film liquid system at varied initial plant densities. Equivalent ratios of
explants per volume media were used for both agar and liquid media.
Initial
density
(plants/L)
40
80
120
200
Significant
Linear
Fit
New plants m-2 wk-1
Multiplication rate
Agar
Hosta sppa.
2.1 ± 0.2
1.7 ± 0.1
1.8 ± 0.1
1.7 ± 0.1
L*
Liquid
Agar
Liquid
3.4 ± 0.2
2.6 ± 0.1
2.7 ± 0.2
2.3 ± 0.2
L***
9±2
12 ± 2
21 ± 3
29 ± 3
L***
20 ± 2
29 ± 3
44 ± 5
55 ± 7
L***
Alocasia macrorrhizab
33
2.1 ± 0.3
3.5 ± 0.4
8±2
13 ± 2
100
1.8 ± 0.2
2.3 ± 0.1
17 ± 4
21 ± 2
165
1.7 ± 0.1
2.4 ± 0.1
24 ± 5
36 ± 4
330
1.3 ± 0.1
1.9 ± 0.1
23 ± 5
45 ± 8
L***
L* Q*
L***
Significant L*
Linear
Fit
a. Data were pooled for three varieties over two, 6-week culture cycles on 1 µM BA. (calculated based on
data from Adelberg 2004).
b. Data were pooled for two media (1 µM BA and 3 µM BA+ 3 µM ancymidol) for a 4-week culture cycle.
33% more media per area shelf space was used in agar jars (calculated from Adelberg and Toler 2004).
In agitated thin-films, Hosta multiplied faster and developed into larger plants than on
agar [22]. Multiplication rate was higher at low plant densities (Table 3). This
phenomenon is more important in thin-film liquid, than agar. Sugar use per vessel
increased with density and more sugar was used in liquid than agar at all densities tested
(40 - 200 plants/L). In Alocasia, Colocasia, Hosta, and Hemerocallis, sugar use was
better correlated to biomass than multiplication rate. Plantlets at harvest were in the
range of 9-18% relative dry weight when sugar is ample. Higher plant densities
produced greater dry matter. However with Alocasia and Colocasia, agitated-liquid
high-density cultures (330 plants/L) have lower residual sugar concentration and lower
relative dry weight in plants at harvest than from agar [23]. Agar cultures were not
depleted of sugar in the range of 33 to 330 explants per litre, but thin-film cultures were.
Supplementing high-density liquid cultures prior to harvest should allow high-density
cultures to obtain higher relative dry matter content and raise soluble solids
concentrations. Greater dry weights of Alocasia, Colocasia and Hosta in liquid are due
to a greater availability of sugar compared to agar [24], as is likely for many other
species.
107
www.taq.ir
J. Adelberg
3. Efficiency in process
3.1. SHOOT MORPHOLOGY FOR CUTTING AND TRANSFER PROCESS
Larger plants are a likely outcome of improved growth in larger vessels from TIS
systems and agitated, thin-films. However, during stage II multiplication, large, wet
plants are more difficult to aseptically transfer and require more space in the culture
vessel. A reasonable approach is to use smaller plants to improve efficiency. Extreme
size reductions of organogenic shoot systems described as meristematic nodule or bud
aggregates have been used to control plant morphology for liquid bioreactor systems
[25]. Growth retrardants that inhibit gibberellin synthesis (ancymidol or paclobutrazol)
were useful in reducing shoot size in cucumber, philodendron, and poplar [26,27,28] as
well as, many of the geophytes described in the previous section. Random mechanized
cutting of bud clusters and bulk inoculation of large vessels of liquid medium during
stage II of micropropagation allowed cost savings of 50% to be predicted [29]. Complex
downstream processing, including individual cutting, sorting and grading, was still
required. A solid stationary phase, albeit on agar or liquid plug systems, was necessary
to develop rooted plantlets. In highly automated attempts to mechanize
microrpropagation, machine vision algorithms, artificial intelligence and robotic
manipulations of tissue have not justified costs.
Manual cutting and re-planting at the hood station is required for virtually all
micropropagation and estimated to be 60% of labour cost [30]. In a hand-cut process for
stage II multiplication with conventional agar media, approximately 7% of time is
required to remove plants, 48% of time is required to cut and 45% of time was required
to re-plant a new vessel [31]. Re-planting gelled media involves repetitive, careful
orientation and spacing each individual bud. In agitated liquid media, bulk transfer of
cut buds during re-planting allows passive spacing and orientation during growth with a
concomitant reduction in technician time at the transfer station. No longer encumbered
by re-planting, the technician may focus entirely on the cutting process. In a commercial
beta-site operation, technicians logged six months of hood time working with a 10-L
Nalgene Biosafe Box (Nalge Nunc International, Rochester, NY, USA) and a bulk
transfer process. Numbers of plants harvested per vessel was the most significant factor
affecting transfer rate when cuts per hour was partitioned by individual technician, plant
variety, media formulation, time of day, day of week and numbers of plants harvested
per vessel [22]. Cutting efficiency increased as plants harvested per vessel increased to
about 100 per vessel. Transfer rate with the Biosafe was low because of excessive size
and an awkward closure system.
During shoot bud division in Stage II old leaves and roots are removed prior to replanting. Nitrogen depletion caused excessive root elongation for birch and orchid
plantlets [32,33]. Preventing tangled root overgrowth by timely harvest schedules is
effective in reducing cutting times. Ancymidol has been used to reduce leaf size of
Hemerocallis, Hosta and ornamental taros - Alocasia and Colocasia with a greater
number or smaller plants grown per vessel [24,34,35]. Ancymidol (0.32 µM) in liquid
cultures of Hemerocallis 'Todd Monroe' with bulk-transfer process decreased plants size
by approximately 50% (FW), increased the numbers or plants per vessel from 60 to 120,
and increased the number of plants cut per hour from 110 to 230 [34]. Ancymidol in
108
www.taq.ir
liquid media also increased sugar uptake and endogenous carbohydrate concentrations,
with varied influences on plant quality in Narcissus, Hemerocallis and Hosta
[24,35,36,37]. Ancymidol and paclobutrazol improve desiccation resistance as part of
an in vitro hardening process for acclimatization [38]. PGR's with lasting downstream
effects may benefit several aspects of a propagation system when correctly integrated.
3.2. SPACE UTILIZATION ON CULTURE SHELF
Round 'baby-food' jars are most frequently used for micropropagation due to their low
cost. Dimensions vary based on market requirements in processed food industries. One
typical vessel, a 180 ml cylindrical baby food-jar has 18 cm2 bottom surface for plant
growth. Eight of these typical vessels in a 4 x 2 arrangement create roughly the same
'footprint' on a culture room shelf as an 11 cm x 27 cm (297 cm2) rectangular vessel
designed for thin-film culture. Yet, the eight jars have a combined interior growth
surface of 144 cm2 (144 cm2 = 4 x 2 x 18 cm2 per jar) that is less than half of the 297
cm2 of the rectangular vessel used for thin-film vessel. Large rectangular vessels create
less void space between vessels on a culture room shelf than larger numbers of smaller
cylindrical jars.
Agar in jars and rectangular thin-films were compared with Hosta (40-200 plants/L)
and Alocasia macrorrhiza over a wider range of densities (33-330 plants/L). As
described in section 3.3., there were higher multiplication rates in liquid than agar, and
the magnitude of this effect was greater at low densities. However, more new plants
(per area shelf space per unit time) were initiated at higher plant densities based on the
greater number plants initially in the vessel. Yields were higher in rectangular thin-film
liquid vessels than round vessels agar-containing medium (Table 3). Yield of Alocasia
in jars levelled-off between 165-330 plants/L, but increased in thin-films liquid over the
entire range of densities tested. Optimization of thin-film system involves low-densities
early in production cycle when rapid increase of plants is most desired. During the peak
production season, high-density cultures would be favoured to obtain greatest output
from a facility with least labour. The large boxes of liquid media permitted the greatest
yields at the highest densities. A second ornamental taro, Colocasia esculenta
'Fontanesii' had similar multiplication rates in agar and liquid but highest yield of new
plants in liquid system (data calculated from [23]). The greater yield of the agitated,
thin-film liquid was likely a combined effect of a) increased surface area for plant
growth within the vessel, and b) larger contact surface of plants and media allowing
greater sugar availability.
3.3. PLANT QUALITY
Hosta from shake-flask culture had greater dry weight than plants from agar. During
subsequent acclimatization, plants from liquid grew faster in greenhouse mist frame and
109
www.taq.ir
J. Adelberg
outdoor nursery [39]. All of the Hosta plants from the density experiment (described in
section 2.3) were successfully acclimatized in the greenhouse. Plants derived from
liquid and agar culture showed comparable vigorous growth to that of greenhouse and
quality was also acceptable.
Plants of Alocasia and Colocasia from the agitated, thin-film liquid system had 2.5
times greater dry weight per plant than from agar (Table 4). Benzyladenine
concentration was raised from 1 µM to 3 µM and ancymidol was added in equimolar
proportion to reduce plant size and problems with tangled plants in transfer. This
resulted in a 45% reduction in dry weight per plant. Plants from all treatments had
greater mean dry weights from liquid than agar at all densities. Greater than 99% of
plants (from 450 sampled of varied sizes) from liquid media acclimatized to greenhouse
and were of acceptable quality. The agitated liquid, thin film system with bulk dump
process allows managers to use higher plant densities while maintaining plant quality.
When compared to agar, this system allowed more and larger plants produced in less
space per unit time with reduced labour.
Table 4. Mean dry weight per plant of two species of ornamental taros after 4 weeks of
culture in agar and agitated, thin film liquid system at different initial plant densities.
Equivalent ratios of explants per volume media was used for both agar and liquid media.
Initial density
Growth medium
(1 µM BA)
Multiplication medium
(3 µM BA + 3 µM Ancymidol)
(plants/L)
Agar
Liquid
Agar
Alocasia macrorrhiza (mg dry weight per plant)
Liquid
33
100
109a ± 22
123 ± 28
195 ± 13
187 ± 49
33 ± 23
26 ± 9
119 ± 19
85 ± 14
167
44 ± 28
142 ± 23
25 ± 10
75 ± 18
330
66 ± 17
93 ± 23
36 ± 5
119 ± 44
Colocasia esculenta 'Fontanesii' (mg dry weight per plant)
33
14 ± 2
31 ± 21
16 ± 2
57 ± 3
100
167
22 ± 7
11 ± 3
161± 42
84 ± 26
21 ± 2
27 ± 3
75 ± 18
43 ± 4
330
23 ± 3
80 ± 7
25 ± 7
50 ± 8
a. Mean dry weight per plant was calculated as the product of biomass per plant and relative dry weight per
vessel from data of Adelberg and Toler, 2004.
4. Vessel and facility design
4.1. PRE-EXISTING OR CUSTOM DESIGNED VESSEL
A vessel needs to be inert, inexpensive and easy to handle. Complete sterilization of all
interior surfaces is essential. Single use vessels sterilized by gamma irradiation or
ethylene oxide are preferred in the biomedical trade but tend to be too expensive for
micropropagation. Vessels need withstand 121oC at 1.2 kg cm2 pressure generated
110
www.taq.ir
during steam sterilization. Translucent materials are necessary to allow light
transmittance into the vessel and an unobstructed view of plant material is important for
quality management by visual inspection. Glass and polycarbonate are the two most
commonly used materials for rigid vessels. Glass is heavy and breaks easily.
Polycarbonate is expensive and becomes clouded with age. Autoclave stable, flexible
film laminates are more difficult to fill with media and tissue. A single preferred
material does not exist. Combining rigid multiple-use, and flexible single-use
components may allow further innovations in vessel construction.
It is desirable to use the fewest parts possible in a vessel system. Each part needs to
be cleaned and inspected during re-use, prior to assembly. Critical surfaces must be
easily accessible and improper decisions made by workers in the dish room impede
successful commercial implementation. Custom fabrication should only be considered
after searching what is available, and what can be easily modified from what is already
available. Work described in this Chapter was first conceived using modifications of the
Nalgene Biosafe, but it was expensive, consisted of 11 parts, and required modification
to allow ventilation and media sampling. It also deformed during steam sterilisation and
was too large to be easily handled at the hood station. However, a mock-up commercial
process with the Biosafe showed value of agitated, thin-films in micropropagation. This
allowed decisions to be made on desired qualities of a custom vessel for agitated, thinfilm culture.
4.2. SIZE AND SHAPE
Rigid vessels are easier to handle than flexible films. The expense of moulding a rigid
vessel dictates considerations of inter-related aspects of process. Economy requires the
fewest custom parts. Thermoforming techniques (injection mould, blow mould, vacuum
mould, etc.) impact cost of the mould and limit choices of size, shape and the precision
of critical surfaces. The mould will cost more than the materials until thousands of units
have been cast. Detailed discussion of plastic fabrication is beyond the scope of this
chapter.
Rectangular vessels were selected for minimal void space and maximized growth
surface for the plants. A base with one longer dimension, allowed a slight pitch to create
a wave capable of immersion of the entire plantlet. Pitch angles ranging from 5-30o
were effective in a vessel with length of 27 cm and width of 10 cm containing 150-250
ml of medium. Length to width ratios greater than 3 are often considered awkward for
handling. The 10 cm base created a large growth surface and a taper to a 6 cm upper
surface made the vessel easier to grip for smaller hands. Vessels were large enough to
allow at approximately 75-150 plants to be harvested per cycle for labour efficiency
[22]. The height of the vessel (10 cm) was determined from other vessels common in
the trade. The side-mounted closure allows greater growth surface to be accessible to a
forceps with advantages in aseptic hood process explained in section 4.3.
Autoclave capacity may limit laboratory throughput. In the US, most autoclaves are
circular bores, horizontally mounted, with flat tray bottoms. A well-designed vessel
should fit most common autoclaves with minimal void space. If vessels are to be
stacked in autoclave, a route for steam penetration within the stacksmust be maintained.
The Liquid Lab Vessel® (Figure 3) fits the Market Forge Sterilmatic STME Autoclave
(Market Forge Industries, Everett MA, USA) in two stacks of four. The fluted top of the
111
www.taq.ir
J. Adelberg
Figure 3. Liquid Lab Vessel® for agitated thin film micropropagation with adhesive
ventilation patches (shown in foreground).
vessel allows steam penetration to media on the bottom surface of the upper vessel layer
during steam sterilization. Back to back arrangement of vessels in the upper layer
accommodates the narrowed width at the top of the autoclave's bore.
Stacking of vessels during storage is facilitated by internally nested, tapered vessels or
collapsed flexible film bags. This convenience was not achieved in the vessel shown.
4.3. CLOSURES AND PORTS
Closures and ports may be made of dissimilar materials from the vessel body. There
must be enough elasticity to allow expansion and contraction during autoclave cycle.
Rigid polycarbonate vessels with softer polypropylene closures are often combined. A
snug interference fit seals by forcing the softer polypropylene cap to conform to the
rigid polycarbonate vessel. For economy, vessels may be moulded to match a preexisting closure. The seal is the most expensive part of the vessel and its length should
be minimized with respect to a maximum growth area. The opening need to be large
enough to allow cut buds be introduced and larger plantlets be removed (disposable
vessels can be cut open at harvest and have much smaller closures). Circular closures
using threaded screw-caps apply uniform pressure on the seal. Thread patterns trap
condensed water and potentially provide refuge for contaminants that could be drawn to
the mouth of the vessel by the screw mechanism when opening. Thread design for
aseptic culture vessels involve fewer concentric rings with greater pitch than those
designed for food containers. The seal should not have broad horizontal surface that
allow condensation to collect.
Gas exchange between the vessel and the ambient environment is necessary to
maintain adequate levels of CO2, O2 and water vapour [40]. A tightly sealed vessel may
not allow adequate ventilation. Loose caps will ventilate the vessel but contamination
112
www.taq.ir
may occur with macroscopic voids. Membrane filters laminated to adhesives and
structural supports may be fixed to openings designed specifically to ventilate the
vessel. Microbes are excluded based on size. Ventilation patches become more cost
effective when larger patches are applied to greater surface areas for growing more
plants in larger vessels. Repeated aseptic sampling of liquid media during the culture
cycle is possible using silicone rubber septa and syringe needles.
4.4. BIOTIC CONTAMINANTS
It is common tissue culture lore that liquid medium is more prone to contamination than
agar. This misstatement is based on reasonable observations. Endogenous contaminants
fastidious to the plant are easier to find suspended in turbid liquids than as cryptic 'white
ghosts' hidden from sight underneath the base of the plantlet embedded in agar.
Generally, bacteria and fungus will grow more quickly in agitated liquids than under
agar medium. Ironically, this property of liquid culture allows a proactive manager
greater lead time to take appropriate action.
Frequently liquid culture involves using larger vessels. More initial explants
increase the chance of contamination as an exponential function of the fraction of plants
that are contaminated. The cost of losing larger batches is higher and so a laboratory's
'base' contamination rate will dictate a reasonable scale of operation. Contaminant
problems introduced in aseptic transfer process are exacerbated by work with larger
vessels in the laminar flow hood. The longer the vessel remains open, the greater the
size of the opening, more frequent or invasive entries, hands or tools crossing over the
entry port, and blocking of laminar flow to the entry port, all increase the chance of
failure with larger vessels. Also, many experiences with larger vessels involve
improvised parts, ill-conceived autoclave packing and ad hoc cooling procedures. These
failures are not due to liquid culture per se, but are problems of larger vessels, itinerant
hardware and protocol.
A process for use of Liquid Lab Vessel® was developed to circumvent
contamination problems. During sub-culture, vessel is placed in the hood so laminar
flow is parallel to the long, linear dimension. A 25 cm forcep is used to remove a
portion of plantlets with the operators' hand shielded from the growth surface by the
vessels slanted, fifth side. Plants should not contact the outer rim of the vessel during
removal. If the plants are too large or entangled, one may consider shorter culture
period or use of ancymidol. Adequate numbers of buds for re-initiation of new vessels
should be cut and stored in sterile, empty jars. Transfer of cut buds to each new vessel
will be made in one motion and only the sterile jar need cross over the entry port. The
size of the entry port in Figure 3 is similar to the size of petri-plate and the time the
vessel remains open during inoculation has been minimized.
4.5. LIGHT AND HEAT
Large, flat transparent surfaces permit unobstructed observation of plants on the bottom
and backside of the culture vessel. Cool white fluorescent light transmitted through the
vessel provides both photosynthetic energy and signals that promote shoot
development. Long tubes provide relatively even distribution of light flux density on the
culture shelf [40]. Light fixtures are typically mounted on the underside of the shelf to
provide downward lighting even though downward lighting does not always provide the
113
www.taq.ir
J. Adelberg
best quality growth [40]. The large, clear bottom surface of Liquid Lab Vessels®
allowed shelves to be reduced to open support frames with light penetration coming
from through the open bottom. Reflectors and canisters were removed from fluorescent
tubes so light would be radially transmitted. This allowed two culture shelf-layers to be
sandwiched between upper and lower lighting layers (Figure 4). Approximately 70% of
the irradiance in the upper shelf came in the downward direction with the other 30%
coming through the filled lower shelf.
Figure 4. Floor to ceiling arrangement of open-frame shelving in 3.7 m culture room.
Similarly, 70% of the irradiance on the lower shelf came in the upward direction
through the frame (with the other 30% coming through the filled upper shelf). The sums
of downward and upward irradiance were equivalent on upper and lower shelves. There
was no difference for multiplication rate, sugar use or appearance of plants in
comparisons between upper and lower shelves during three years of pilot scale process
with thousands of vessels.
Electricity is approximately 5% of the cost of goods in a commercial lab [30].
Working with a 3.7 m shelving stack allowed 10 shelves (5 pairs of upper and lower) to
utilize 7 rows of light fixtures, not 10. Lighting the culture room is about 65% of the
electricity cost, and cooling those lights is another 25% of the electric cost. The 30%
reduction of light fixtures is therefore significant. Theoretically, the number of lights
would be reduced by 50% as the stacks get taller, but this creates man-motion and
worker safety as constraints.
Removing heat trapped in tightly filled solid shelves limit how closely shelving may
be arranged in a traditional vertical stack of shelves. Tilted, open frames did not appear
114
www.taq.ir
to trap heat, even when packed with vessels. The rocking motion dissipated any 'hot
pockets' with a bellow-type motion. Tight vertical packing of shelf-pairs allows stacked
planar growth surface areas to be optimized in the volume of space under environmental
control.
Three-dimensional volumetric optimization in full immersion bioreactors is
theoretically the most efficient way to grow plant cells. As the organism develops
polarity, aerated shoots fixed in gaseous phase become more important to plant quality.
Optimization of two dimensional growth surfaces for nutrient exchange, with an
adequate aerial environment is necessary for micropropagation of shoots and plants of
most species. Latter stages of somatic embryo conversion may similarly benefit from
these approaches. If a system is to be readily used, it must conform to the human
environment - simple, economic and robust. In this current iteration, the bioreactor has
been simplified to a vessel that is placed on the shelf without mechanical linkages to
pumps and motors. Unit size for plant-handling was dictated by the technician.
Managers realize a scale-up factor that allows more active monitoring of process.
Reasonably sized factorial experiments may rapidly determine optimization of
genotype, PGR or nutrient-use scenarios. Values added to the young plant by enhanced
transfer of nutrients can be delivered in the market competitively with plants produced
on agar.
Disclaimer
The use of trade names does not imply product endorsement by the author, or Clemson
University
References
[1] Adelberg, J.; and Simpson, E.P. (2004) Intermittent immersion vessel apparatus and process for plant
propagation. US Patent 6,753,178 B2.
[2] Smith, M.A.L.; and Spomer, L. (1995) Vessels, gels, liquid media and support systems. In: AitkenChristie, J.; Kozai, T. and Smith, M.A.L. (Eds.) Automation and environmental control in plant tissue
culture. Kluwer Academic Publishers, Dordrecht, The Netherlands; pp. 371-405.
[3] Williams, R.R. (1995) The chemical microenvironment. In: Aitken-Christie, J.; Kozai, T. and Smith,
M.A.L. (Eds.) Automation and environmental control in plant tissue culture. Kluwer Academic
Publishers, Dordrecht, The Netherlands; pp. 405-440.
[4] Leifert, C.; Murphy, K.P. and Lumsden, P.J. (1995) Mineral and carbohydrate nutrition of plant cell and
tissue cultures. CRC Crit. Rev. Plant Sci. 14: 83-109.
[5] Williams, R.R. (1993) Mineral nutrition in vitro - a mechanistic approach. Austr. J. Bot. 41: 237-251.
[6] Pryce, S.; Lumsden, P.J.; Berger, F.; Nicholas, J.R. and Leifert, C. (1994) Effects of plant density and
macornutrient nutrition on Iris shoot cultures. In: Lumdsen, P.J.; Nicholas, J.R. and Davies, W.J. (Eds.)
Physiology, Growth and Development of Plants in Culture. Kluwer Academic Publishers, Dordrecht, The
[7] Leifert, C.; Lumsden, P.J.; Pryce, S. and Murphy, K.P. (1991) Effects of mineral nutrition on the growth
of tissue cultured plants. In: Goulding, K.H. (Ed.) Horticultural Exploitation of recent biological
developments, Proceedings of the Institute of Horticulture, Sep. 1991; pp. 43-57.
[8] Ibaraki,Y. and Kurata, K. (1993) Comparison of culture methods from the viewpoint of nutrient
movement. ASAE paper no. 934049, Presented Spokane WA; June 1993.
115
www.taq.ir
J. Adelberg
[9] Ibaraki,Y. and Kurata, K (1998) Relationship between water content of Cymbidium protocorm-like body
and growth. In L.F.M (ed.) Crop models in protected cultivation. Acta Hort. 456: 61-66.
[10] Ramage, C.M. and Williams, R.R. (2003) Mineral uptake in tobacco leaf discs during different
developmental stages of shoot organogenesis. Plant Cell Rep. 21: 1047-1053.
[11] Desamero, N.; Adelberg, J.; Hale, A.; Young, R. and Rhodes. B. (1993) Nutrient utilization in
liquid/membrane system for watermelon micropropagation. Plant Cell Tissue Org. Cult. 33: 265-271.
[12] Curtis, W. (1999) Achieving economic feasibility for moderate value food and flavour additives: a
perspective on productivity and proposal for production technology cost reduction. In: Fu, T.J.; Sing, G.
and Curtis, W. (Eds.), Plant Cell and Tissue Culture for Production of Food Ingredients. Kluwer
Academic/ Plenum Publ., New York; pp. 225-236.
[13] Gollagunta, V.; Adelberg, J.; Rajapakse, N. and Rieck, J. (2004) Media composition affects carbohydrate
status and quality of Hosta tokudama Tratt. 'Newberry Gold' micropropagules during low temperature
storage. Plant Cell Tissue Org. Cult. 77: 125-131.
[14] Gollagunta, V.; Adelberg, J.; Rajapakse, N. and Rieck, J. (2005) Sucrose in storage media and cultivar
affects post-storage re-growth of in vitro Hosta propagules. Plant Cell Tissue Org. Cult. (In press).
[15] Lian, M.; Chakrabarty, D. and Paek, K.Y. (2002) Growth and uptake of sucrose and minerals by bulblets
of lilium oriental hybrid 'Casablanca' during bioreactor culture. J. Hort. Sci. Biotechol. 77: 253-257.
[16] Kim, E.K.; Hahn, E.J.; Murthy, H.N. and Paek, K.Y. (2003) High frequency shoot multiplication and
bulbet formation of garlic in liquid cultures. Plant Cell Tissue Org. Cult. 73: 231-236.
[17] Ziv, M. and Shemesh, D. (1996) Propagation and tuberization of potato bud clusters from bioreactor
culture. In Vitro Cell. Dev. Biol.-Plant 32: 31-36.
[18] Salvi, N.D.; George, L. and Eapen, S. (2002) Mircropropagation and field evaluation of micropropagated
plants of tumeric. Plant Cell Tissue and Org. Cult. 68: 143-151.
[19] Zhou, S.; He, Y. K. and Li, S. (1999) Induction and characterization of in vitro corms on diploid taro.
Plant Cell Tissue Org. Cult. 57:173-178.
[20] Etienne, E. and Berthouly, M. (2002) Temporary immersion systems in plant micropropagation. Plant
Cell Tissue Org. Cult. 69: 215-231.
[21] Escalona, M.; Samson, G.; Borroto, C. and Desjardins, Y. (2003) Physiology of effects of temporary
immersion bioreactors on micropropagated pineapple plantlets. In Vitro Cell. Dev. Biol-Plant.39: 651656.
[22] Adelberg, J. (2004) Efficiency in thin-film liquid system for micropropagation of Hosta. Plant Cell
Tissue Org. Cult. (In press).
[23] Adelberg, J. and Toler, J. (2004) Comparison of agar and an agitated, thin-film liquid system for
micropropagation of ornamental elephant ears. Hort. Sci. 39: 1088-1092.
[24] Adelberg, J. (2004) Plant growth and sugar utilization in an agitated, thin film liquid system for
micropropagation. In Vitro Cell. Dev. Biol.-Plant. 40: 245-250.
[25] Ziv, M. (1999) Organogenic plant regeneration in bioreactors. In: Altman, A.; Ziv, M. and Izhar, S.
(Eds.) Plant Biotechnology and In Vitro Biology in the 21st Century. Kluwer Academic Publsihers,
[26] Ziv, M. (1992) The use of growth retrardants for the regulation and acclimatization of in vitro plants. In:
Karsen, C.; Van Loon, L. and Vregdenhil, D. (Eds.) Progress in plant growth and regulation. Kluwer
Academic Publishers, Dordrecht, The Netherlands; pp. 809-817.
[27] Ziv, M.. and Ariel, T. 1991. Bud proliferation and plant regeneration in liquid cultured Philodendron
treated with ancymidol and paclobutrazol. Plant Growth Regul. 10: 53-57.
[28] Vincour, B.; Carmi, T.; Altman, A. and Ziv, M. (2000) Enhanced bud regeneration in aspen (Populus
tremula L.) roots cultured in liquid media. Plant Cell Rep.19: 1146-1154.
[29] Gross, A. and Levin, R. (1999) Design consideration for mechanized micropropagation laboratroy. In:
Altman, A.; Ziv, M. and Izhar, S. (Eds.) Plant Biotechnology and In Vitro Biology in the 21st Century.
Kluwer Academic Publsihers, Dordrecht, The Netherlands; pp. 637-642.
[30] Chu, I. (1995) Economic analysis of automated micropropagation. In: Aitken-Christie, J.; Kozai, T.and
Smith,M.A.L. (Eds.) Automation and environmental control in plant tissue culture. Kluwer Academic
Publishers, Dordrecht, Netherlands; pp. 19-28.
[31] Alper, Y.; Young, R.; Adelberg. J. and Rhodes, B. (1994) Mass handling of watermelon microcuttings.
Trans. Amer. Soc. Ag. Eng. 37: 1337-1343.
[32] Adelberg, J.; Desamero, N.; Hale, A. and Young, R. (1997) Long-term nutrient and water use during
micropropagation of Cattleya orchid on liquid/membrane system. Plant Cell Tissue Org. Cult. 48: 1-7.
116
www.taq.ir
[33] McDonald, A.J.S. (1994) Nutrient supply and plant growth. In: Lumdsen, P.J.; Nicholas, J.R. and
Davies, W.J. (Eds.) Physiology, Growth and Development of Plants in Culture. Kluwer Academic
Publishers, Dordrecht, The Netherlands; pp. 47-57.
[34] Maki, S.; Delgado, M. and Adelberg, J. (2005) Time course study of anycmidol on microporpagated
Hosta. Hort. Sci. (In press).
[35] Adelberg, J.; Delgado, M. and Tomkins, J. (2005) Ancymidol and liquid media improved
micropropagation of Hemerocallis cv. Todd Monroe on the 'rocker' thin-film bioreactor. J. Hort.
Biotechnol. (In press).
[36] Chen, J. and Ziv, M. (2001) The effect of ancymidol on hyperhydricity, regeneration, starch and
antioxidant enzymatic activities in liquid-cultured Narcissus. Plant Cell Rep. 20: 22-27.
[37] Chen, J. and Ziv, M. (2003) Carbohydrate, metabolic, and osmotic changes in scaled-up liquid cultures
of Narcissus leaves. In Vitro Cell. Dev. Biol-Plant 39: 645-650.
[38] Ziv, M. (1995) In vitro acclimatization. In: Aitken-Christie, J.; Kozai, T. and Smith, M.A.L. (Eds.)
Automation and environmental control in plant tissue culture. Kluwer Academic Publishers, Dordecht,
[39] Adelberg J.; Kroggel, M. and Toler, J. (2000) Greenhouse and nursery growth of micropropagated
Hostas from liquid culture. Hort. Tech. 10: 754-757.
[40] Fujiwura, K. and Kozai, T. (1995) Physical microenvironment and its effect. In: Aitken-Christie, J.;
Kozai, T. and Smith, M.A.L. (Eds.) Automation and environmental control in plant tissue culture.
Kluwer Academic Publisher, Dordrecht, The Netherlands; pp. 319-369.
117
www.taq.ir
DESIGN, DEVELOPMENT, AND APPLICATIONS OF MIST BIOREACTORS
FOR MICROPROPAGATION AND HAIRY ROOT CULTURE
MELISSA J. TOWLER1, YOOJEONG KIM2, BARBARA E.
WYSLOUZIL3, MELANIE J. CORRELL4, AND PAMELA J.
WEATHERS1
1
Department of Biology/Biotechnology, Worcester Polytechnic Institute,
Worcester, MA,01609,USA - Fax: 508-831-5936 -Email:
[email protected]
2
Department of Chemical Engineering, Worcester Polytechnic Institute,
Worcester, MA01609,USA
3
Department of Chemical and Biomolecular Engineering, The Ohio State
University, Ohio, USA – Fax: 614-292-3769
4
Agricultural and Biological Engineering Department, University of
Florida, Gainesville, FL 32611, USA-Fax: 352-392-4092
1. Introduction
Aeroponic technology has been used extensively to study biological phenomena in
plants including drought stress, symbiotic relationships, mycorrhizal associations,
disease effects, mineral nutrition, overall plant morphology and physiology [1], and
some work has also been completed with animal tissue culture [2]. Aeroponics offers
many advantages to whole plant growth because of the enhanced gas exchange that is
provided. Here we focus on the use of aeroponics (nutrient mists) for in vitro culture of
differentiated tissue, in plant micropropagation, and in the culture of transformed
(hairy) roots for secondary metabolite production.
There are two main categories of bioreactors: liquid-phase and gas-phase reactors
[3]. In liquid-phase reactors, the tissue is immersed in the medium. Therefore, one of
the biggest challenges in a liquid-phase culture is delivering oxygen to the submerged
tissues due to low gas solubility. In gas-phase reactors (which include nutrient mist
culture), the biomass is exposed to air or a gas mixture and nutrients are delivered as
droplets. Droplet sizes can range from 0.01-10 µm for mists, 1-100 µm for fogs, and 10103 µm for sprays [4]. The mass transfer limitation, especially of oxygen, can be
significantly reduced or eliminated by using a gas-phase culture system [5].
119
© 2008 Springer.
www.taq.ir
M.J. Towler, Y. Kim, B.E. Wyslouzil, M.J. Correll and P.J. Weathers
2. Mist reactor configurations
The original design of aeroponics systems dispersed nutrient medium via spray nozzles
that required compressed gas and were prone to clogging by medium salts [1], while
later mist reactors used submerged ultrasonic transducers. In the early mist reactors
(Figures 1A and 1B), the ultrasonic transducer was in direct contact with nutrient
medium salts and had to be autoclaved, considerably shortening the life of the
transducer [6-8]. Buer et al. [8] fabricated an acoustically transparent polyurethane
window to isolate the medium from the transducer (Figure 1C) but making the windows
was difficult, time consuming, and the starting materials were expensive. Chatterjee et
al. [9] replaced the custom window with an inexpensive, commercially available
polypropylene container (Figure 2) and this design was successfully used for both hairy
root [9] and micropropagation studies [10-12]. Similarly, Bais et al. [13] used a
polycarbonate GA-7 vessel. The nutrient mist system currently used by Weathers et al.
[5] (Figure 3) has an acoustic window consisting of a thin sheet that has a higher
temperature tolerance than polypropylene and can also be incorporated into a reactor of
almost any size or shape. The designs of the mist reactor configuration have evolved as
the applications of these systems have become more varied.
Figure 1. Three types of ultrasonic mist reactors: the mist generator and the growth
chamber are in separate vessels (A) mist generator and growth chamber are in the same
vessel (B) and the transducer is separated from autoclaved components by an acoustic
window (C). Direction of mist movement is indicated by arrows.
120
www.taq.ir
www.taq.ir
Design, development, and applications of mist bioreactors for micropropagation and hairy root culture
Figure 2. Acoustic window mist reactor; A, mist generator; B, micropropagation
chamber; C, media reservoir; 1, polypropylene mist chamber; 2, nutrient medium level;
3, Holmes® humidifier base; 4, ultrasonic transducer; 5, coalescer; 6, one-way valve; 7,
micropropagation chamber; 8, plant platform; 9, gas sampling port; 10, chamber
supports; P, peristaltic pump used for pumping medium to mist chamber.
Figure 3. Two types of gas-phase bioreactors for hairy root culture. Top, trickle bed
reactor. Bottom, nutrient mist reactor.
121
www.taq.ir
3. Mist reactors for micropropagation
Worldwide, an estimated one billion plants per year are produced by micropropagation
[14]. In the micropropagation scheme (Figure 4), [15,16], stage 0 is the selection of the
donor plant, and may involve genetic testing and disease indexing. In stage I, the
explant (generally the shoot tip) is isolated and disinfected and sterile culture is initiated
on an appropriate nourishing medium. Multiplication of the explant occurs in stage II,
usually via exogenous hormonal stimulation of branching, with subcultures performed
as needed. In stage III, the shoots are stimulated to produce roots by altering the
hormone content of the medium. Sometimes rooting is initiated instead in conjunction
with stage IV (acclimatization) to prevent damage to the fragile newly formed roots
during transfer. While roots that develop in vitro are often considered non-functional,
for some plants the presence of in vitro roots at the time of transplanting may have
beneficial effects on the plant's water status [17]. Acclimatization (stage IV) may take
weeks as the plant makes the transition to the non-sterile environment at lower relative
humidity, and greater light intensity rates. The high relative humidity of the in vitro
culture causes changes to the structure of the shoot’s cuticle, wax deposits, stomata and
mesophyll cells, subsequently inhibiting photosynthesis. Therefore, the plants must
"learn" how to photosynthesize [18]. The final stage, stage V, involves verifying the
status of the plant with respect to its genetic integrity and disease-free condition.
An important advantage of gas-phase systems such as a nutrient mist bioreactor
(mist reactor) when used for micropropagation is the potential for precise control of the
gas composition and relative humidity surrounding the plants because these parameters
can significantly affect multiplication rates, rooting, and acclimatization [19,20]. Design
and development of an effective and inexpensive mist reactor for micropropagation,
however, presents engineering challenges unique to this application. A summary of
studies using mist reactors for micropropagation is provided in Table 1.
Typical in vitro micropropagation environments have high relative humidity (95100% RH), low light intensity (30-75 Pmol m-2s-1), and large fluctuations in CO2 [21].
These conditions can contribute to increased hyperhydration [22], reduced
photosynthetic ability [23], or increased transpiration [24] in plants when compared to
field-grown specimens. The presence of supplemental sucrose in the growth media to
compensate for decreased photosynthesis can also reduce fixation of CO2. Further
deficiencies of CO2 result from the culture chamber, which is sealed in order to
maintain the sterility of the carbon-rich media, which also leads to poor gas exchange
between the tissue and the outside atmosphere.
The gaseous composition of the headspace within tissue culture vessels is a major
factor influencing plant growth and development in vitro [25], and depending on the
volume of the vessel and the extent of ventilation, is composed mainly of nitrogen,
oxygen, carbon dioxide, and may contain ethylene, ethanol, acetaldehyde, and other
hydrocarbons [26]. One of the main problems encountered by plants in an in vitro
environment is hyperhydration, which is caused by the inadequate headspace conditions
in the culture vessels typically used for micropropagation. Hyperhydration results in
poor plant development in vitro and, later, ex vitro [26]. Plants that are hyperhydrated
often do not survive outside of their protected in vitro environment [27]. Using a mist
122
www.taq.ir
reactor, Correll et al. [10] were able to reduce hyperhydration in Dianthus caryophyllus
plants by altering the mist feed rate anddutycycle
Table 1. Summary of micropropagation mist reactor studies.
Species
Inoculum
Artemisia
Shoots
Asparagus
Shoots
Asparagus
Shoots
Brassica
Anthers
Capsicum
cell
suspension
nodal explants
Cinchona
Cordyline
Daucus
Daucus
Daucus
shooting
tissue
Callus and
shoots
Shootlets
Dianthus
embryogenic
callus
node cuttings
Dianthus
node cuttings
Dianthus
node cuttings
Dianthus
node cuttings
Ficus
callus
w/shooting
meristems
nodal explants
Lycopersicon
Musa
Nephrolepis
shooting
tissue
Shoots
Nephrolepis
Shoots
Solanum
nodal explants
Solanum
nodal explants
Type of
Mist
system
submerged
ultrasonics
submerged
ultrasonics
submerged
ultrasonics
spray
reactor
spray
reactor
spray
reactor
submerged
ultrasonics
spray
reactor
submerged
ultrasonics
submerged
ultrasonics
acoustic
window1
acoustic
window1
acoustic
window1
acoustic
window1
spray
reactor
spray
reactor
submerged
ultrasonics
submerged
ultrasonics
acoustic
window2
modified
Mistifier™
submerged
ultrasonics
Main results
higher biomass and artemisinin
than liquid reactors
doubled root and shoot initiation
and elongation
higher root and shoot initiation and
elongation
increased regeneration versus agar
fully developed plants after 10
weeks
increased shooting; 20% higher
FW weight than agar
higher shoot production versus agar
3.5x increase in net weight
compared to agar plates
induction of asexual embryoids,
not in liquid or agar
more somatic embryos than agar;
none in liquid controls
growth comparable to test tubes; 2x
less hyperhydration
hyperhydration reduced by misting
scheme
higher ex vitro survival than GA7
culture boxes
hyperhydration reduced by higher
light and CO2
increase in shooting
Reference
[33]
[35]
[36]
[34]
[39]
[34]
[37]
[34]
[6]
[6]
[9]
[10]
[11]
[12]
[34]
higher shoot production versus agar
growth comparable to submerged
ultrasonics and plates
growth comparable to controls
98% of inocula formed tubers
[34]
[37]
[37]
[8]
[32]
[38]
1, polypropylene; 2, Conap's EN6
123
www.taq.ir
Figure 4. Stages of micropropagation.
Light intensity, CO2, and humidity also affect hyperhydration, and the latter two
conditions can be altered using mist reactors [11,12]. CO2 enrichment has been shown
to promote net photosynthesis and prepare plants for ex vitro acclimatization [28] and
may significantly reduce the acclimatization period [29,30]. Increased CO2 levels
decreased hyperhydration in D. caryophyllus plants grown in the mist reactor [12], but
only when used in conjunction with higher light intensity. Taken together, these studies
show that hyperhydration can be reduced or eliminated using a mist reactor where gas
content is regulated.
Acclimatization accounts for approximately 30% of the total production cost of
micropropagation [14]. Correll and Weathers [11] used a mist reactor to grow and
acclimatize carnation plants in vitro without using ex vitro acclimatization techniques,
which are expensive, time-consuming, and labour-intensive [14,31]. Ex vitro plant
survival rates were higher for plants grown in the mist reactor (91% survival) using the
acclimatization protocol described in Correll and Weathers [11] versus a conventional
propagation system (GA-7 culture boxes) that only had a 50% survival rate.
Multiple studies have shown that using the mist reactor in its various configurations
promoted equivalent or better growth of plant inocula compared to traditional controls
[8,9,32-34], increased shooting [34-37], increased formation of somatic embryos [6]
and microtubers [38], and yielded higher rates of regeneration [34,39].
It should also be noted that there appears to be an unusual pattern to the spreading of
contamination through the mist reactor system. While contamination is always a concern
for in vitro systems due to the high sugar content of the medium and the fragile nature of
the plant tissue, recent observations by Sharaf-Eldin and Weathers (unpublished)
124
www.taq.ir
suggest that areas of contamination that develop in the mist reactor growth chamber
remain relatively isolated and progress more slowly than in liquid or semi-solid media.
This phenomenon is presently under investigation.
Although there are many challenges that face the micropropagation industry, the
most prevalent is the cost and time associated with labour. Much of the industry relies
on low-wage workers from underdeveloped countries for their workforce and the
economic and political instability of these countries threatens the success of this
industry. The manual tasks of cutting, transplanting, and acclimatizing plant tissues are
slow and increase the rates of contamination, thereby increasing loss in product and
overall costs. Automation of these steps could decrease production time, lessen
contamination rates, and reduce labour demands. Honda et al. [40] described at length
an image analysis system for robotics-assisted automated selection of plant tissue in
large-scale micropropagation. A mist reactor offers the potential for automating several
other stages in micropropagation and combining shoot and root production with
acclimatization [11].
4. Mist reactors for hairy root culture
A number of valuable pharmaceuticals, flavours, dyes, oils, and resins are plant-derived
secondary metabolites. Since secondary metabolites are usually produced by specialized
cells and/or at distinct developmental stages [41], plant cell suspension cultures are not
usually practical sources of these chemicals. Hairy root cultures can have the same or
greater biosynthetic capacity for secondary metabolite production compared to their
mother plants [42,43]. Indeed, hairy roots have been considered potential production
sources for important secondary metabolites [44]. A summary of studies using hairy
roots in mist reactors is provided in Table 2. In nearly all cases, hairy root growth in
mist reactors was as good as or better than liquid-phase cultures.
Secondary metabolism of hairy roots grown in various bioreactors has been recently
reviewed by Kim et al. [3]. Kim et al. [45] noted a 3-fold increase in artemisinin
accumulation in mist reactors, and subsequently, Souret et al. [46] provided a further
analysis when they compared the expression levels of four key terpenoid biosynthetic
genes in A. annua hairy roots grown in mist reactors versus liquid-phase systems after.
Although there was notable heterogeneity in terpenoid gene expression, the differences
could not be attributed directly to one single factor and were likely the result of complex
interactions of multiple factors including oxygen status, presence or absence of light,
culture age, and tissue location within the growth chamber of the bioreactor. Bais et al.
[13] and Palazon et al. [47] likewise noted alterations in secondary metabolite content
when hairy roots of Cichorium and Panax, respectively, were grown in mist reactors.
Several hairy root lines can develop mature chloroplasts capable of photosynthesis
[48], and these green roots have different metabolic capabilities compared to non-green
roots, although response to light is not necessarily dependent on whether the roots turn
visibly green. In addition, light can have a significant effect on growth of hairy roots
[49] and many enzymes in the biosynthetic pathways for secondary metabolites are
regulated by light [3]. However, delivery of light into a bioreactor, especially one that is
densely packed with roots, is problematic. Interestingly, the roots themselves may have
light-guiding properties [50,51]. A. annua hairy roots were able to transmit light from a
125
www.taq.ir
helium-neon laser through the interior of the root (Weathers and Swartzlander,
unpublished), indicating that roots may have the ability to function as leaky optical
fibers.
Table 2 Summary of hairy root mist reactor studies.
Species
System
Main results
Reference
1
Artemisia
Artemisia
Artemisia
Artemisia
Artemisia
Artemisia
Artemisia
Beta
Carthamus
Cichorium
Datura
acoustic window
mist reactor
submerged
ultrasonics
acoustic window2
mist reactor
acoustic window2
mist reactor
acoustic window2
mist reactor
acoustic window2
mist reactor
acoustic window2
mist reactor
submerged
ultrasonic
submerged
ultrasonics
acoustic window3
mist reactor
droplet reactor
Fragaria
Hyoscyamus
Nicotiana
hybrid
submerged/droplet
reactor
mist reactor
spray reactor
spray reactor
Panax
spray reactor
Datura
growth comparable to flasks and plates
modified inner-loop reactor growth comparable
to flasks
no O2 limitation, but 50% less biomass than
liquid systems
[8]
[69]
[5]
altered branching rate versus flasks
[61]
3 x higher artemisinin content than bubble
column
[45]
growth comparable to bubble column
[55]
altered terpenoid gene expression versus flasks
[46]
growth comparable to flasks
[73]
growth comparable to flasks; 15% faster than
airlift reactor
higher biomass and esculin content than bubble
column
1.6 x lower doubling time than submerged
cultures
successful large-scale (500 L) culture
biomass yield higher than droplet bioreactor
growth comparable to shake flask
50% lower doubling time than flasks
altered ginsenoside pattern versus native
rhizome
[60]
[13]
[74]
[57]
[75]
[52]
[76,77]
[47]
1, Conap's EN6; 2, Teflon; 3, polycarbonate
The morphological characteristics of hairy roots demand special consideration with
regards to bioreactor design. The mist reactor provides a low-shear environment for
growing hairy roots and reduces gas-exchange limitations normally found in liquidphase bioreactors. Studies by McKelvey et al. [52] suggested that roots are more
capable of compensating for poor liquid dispersion than for poor gas dispersion within
reactor systems [53]. An economically viable production scheme depends in part on the
ability to attain a high biomass density. The maximum root tissue concentration that can
be achieved is dependent on the delivery of oxygen and other nutrients into the dense
matrix [54]. Gas-phase reactors such as the mist reactor can virtually eliminate any
oxygen deficiency in dense root beds [5]. Kim et al. [55], however, noted that the
availability of non-gaseous nutrients may be a concern; i.e. gas dispersion is improved
at the expense of liquid dispersion. Furthermore, it is difficult to uniformly distribute
126
www.taq.ir
roots in the growth chamber of a gas-phase reactor without manual loading [3]. Several
groups [44,55-57] circumvented this issue with hybrid liquid and gas-phase reactors,
which were first operated as liquid-phase systems to allow the roots to circulate,
distribute, and/or attach to immobilization points. Gas-phase operation could then be
initiated as desired, usually when the liquid-phase reactor was no longer effective at
supporting root growth due to limitations in nutrient delivery to the dense root beds.
Towler and Weathers [58] have also described a method by which roots may be quickly
attached to a mesh support, thereby allowing mist mode to commence shortly after
inoculation.
The gas phase surrounding tissues also plays a key role in the culture and secondary
metabolite productivity of hairy roots (see review by Kim et al. [3]). One of the major
advantages of the mist reactor is the ability to alter the gas composition. Oxygen is
essential for respiration and thus, the growth of roots. To assess the response of hairy
roots to altered levels of oxygen in mist reactors, alcohol dehydrogenase (ADH) mRNA,
an indicator of oxygen stress, was measured in A. annua hairy roots. Comparison of
ADH mRNA expression in both shake flasks and bubble column reactors to mist
reactors indicated that the mist-grown roots were not oxygen limited [5]. Roots grown
in the mist reactor to a density of about 37% (v/v) had no detectable expression of ADH
[59], whereas ADH mRNA was detected in roots from the bubble column at packing
densities as low as 6% v/v [5]. Roots grown in the bubble column reactor, however, had
higher dry mass compared to those harvested from the mist reactor. This unexpected
result may be explained through modelling of mist deposition dynamics.
In addition to oxygen, carbon dioxide also affects the growth of hairy roots. CO2enriched nutrient mist cultures of Carthamus tinctorius and Beta vulgaris hairy roots
showed increased growth versus control cultures that were fed ambient air [60].
However, a similar effect was not observed in hairy roots of Artemisia annua. When
roots were provided mist enriched with 1% CO2, growth was not significantly different
than that of roots grown in ambient air [61], although visually the roots appeared much
healthier and there was a change in the branching rate. Kim et al. [55] also noted similar
results where the biomass accumulation was similar between root cultures grown in
ambient air and those supplemented with 0.5% CO2. It is possible that perhaps the
optimum level of CO2 enrichment for A. annua hairy roots was not provided to these
cultures, particularly considering that the response of roots to CO2 can vary depending
on species and growth environment [1,60].
Ethylene accumulation may also be involved in regulating biomass and secondary
metabolite production. Although all plant tissues can both produce and absorb the
gaseous phytohormone ethylene, which has profound effects on growth, development,
and even the production of secondary metabolites [62], some species of plants may
produce more ethylene than others. Indeed, Biondi et al. [63] showed that hairy roots of
Hyoscyamus muticus produced 3 times more ethylene than untransformed roots, and
growth of A. annua hairy roots was significantly reduced by ethylene [64]. Sung and
Huang [65] showed that hairy roots of Stizolobium hassjoo had lower biomass and
produced lower levels of secondary metabolites when ethylene was allowed to
accumulate in the headspace of the culture vessel. Recently, we also observed that
ethylene, provided as ethephon, significantly inhibited both growth and artemisinin
production in A. annua hairy roots [64]. Considering that ethylene production is
127
www.taq.ir
inhibited by CO2, it is possible that the stimulation in root growth by higher levels of
CO2 is the result of inhibition of ethylene biosynthesis. Designs in reactors that scrub
ethylene from the gas phase may further improve hairy root growth and promote
secondary metabolite production.
5. Mist deposition modelling
Droplet transport and deposition in a bed of hairy roots may limit growth if an adequate
supply of nutrients does not reach the surface of all roots. Consequently, mist deposition
is a key step in the mass transfer of nutrients to the roots in a mist reactor [66]. The
standard aerosol deposition model for fibrous filters was applied to mist deposition in
hairy root beds by Wyslouzil et al. [66]. The ideal filter has evenly distributed fibres
that lie perpendicular to the flow. Though root beds have regions of high and low
packing density and grow in all directions, the model can still be used to study the
qualitative trends of mist deposition behaviour. When the model was tested on root beds
that had been manually packed to Į = 0.5 (Į = volume fraction occupied by roots), it
was found to correspond well to experimental data as long as the Reynolds number
(Re), based on the root diameter, was <10. The Reynolds number characterizes the
relative importance of inertial and viscous forces, and for filtration problems:
Re = ȡ Uo DR / µg
(1)
where, ȡ and µg are the density and viscosity of the carrier gas, DR is the diameter of the
root, and Uo is the gas velocity in the root bed. In terms of the number of droplets
captured, the efficiency (ȘB) of the root bed is a function of the particle diameter (DP)
and is equal to:
ȘB = 1 - exp [-4 L Į ȘC / (ʌDR (1-Į))]
(2)
where, L is the length of the root bed and:
ȘC = 1 - (1 - ȘIMP +
I NT)
x (1 - ȘD),
(3)
the combined capture efficiency due to impaction, interception, and diffusion,
respectively. Determining ȘIMP + INT involves solving two nonlinear equations [67], and
ȘD may be calculated [68]. The overall mass deposition efficiency (ȘOM) of the root bed
is the product of the root bed efficiency ȘB (DPi) and the mass fraction m (DPi) of mist
particles of diameter DPi summed over the aerosol size distribution data:
ȘOM = Ȉi ȘB (DPi) × m (DPi)
(4)
Typical mist particle size data were obtained experimentally by Wyslouzil et al. [66].
The amount of medium captured by the roots (Vdep) in mL per day is:
Vdep = 24 Ȧ x QL x ȘOM
(5)
128
www.taq.ir
where 24 is the conversion factor from hours to days, Z is the duty cycle in minutes per
hour, and QL is the medium flow rate in mL per minute while misting is occurring. The
amount of medium required to support the growth of roots (Vreq) depends on: the
density of the roots ȡFW (grams fresh weight per mL), the dry weight / fresh weight ratio
(DW/FW), the specific growth rate µ (day-1), the nutrient concentration in the medium
Cs (g per L), the apparent biomass yield of the growth-limiting nutrient YX/S (g DW
biomass per g nutrient consumed), the working volume of the reactor V (L), and
packing fraction Į. The expression for Vreq is:
Vreq = 106ȡFW x DW/FW x µ / CS x 1/YX/S x V x Į.
(6)
The growth-limiting nutrient is assumed to be sugar. Clearly, Vdep must be equal to or
greater than Vreq in order to maintain a desired growth rate µ.
Kim et al. [55] applied the model to A. annua hairy roots grown in the nutrient mist
bioreactor, and it suggested that growth was limited by insufficient nutrient availability.
This hypothesis has been tested in several ways (Towler, unpublished results). Since
Vdep is a function of the packing fraction (Į), increasing Į should increase Vdep and thus
support a higher growth rate by allowing more nutrients to be captured by the roots. To
test this hypothesis, the nutrient mist bioreactor described by Weathers et al. [5] was
modified whereby the growth chamber was replaced with a much smaller (~45 mL
volume, ~30 mm diameter) cylinder into which roots were manually inoculated at an
initial packing fraction of 0.29. The system was then immediately run in mist mode
rather than as a hybrid liquid- and gas-phase reactor. While Kim et al. [55] commenced
mist mode at packing fractions that were at most 0.05 and observed an average specific
growth rate of 0.07 day-1, the average growth rate in the modified mist reactor was 0.12
day-1 for a 6-day period. Due to the disparity in culture times and other operating
conditions, direct comparison between these systems is difficult; however, roots grown
in the modified mist reactor had higher growth rates compared to those obtained by Kim
et al. [55], thereby supporting the hypothesis that initial inoculum density influences
subsequent growth in mist reactors.
Alternatively, since Vreq is inversely proportional to the concentration of the limiting
nutrient CS, increasing CS should decrease Vreq. Using the smaller modified mist reactor
previously described, A. annua hairy roots were fed to the medium containing either 3%
or 5% sucrose. After 6 days, roots grown with 5% sucrose had a significantly higher
specific growth rate compared to roots grown in 3% sucrose (0.18 days-1 and 0.12 days-1
for 5% and 3% sucrose, respectively). Studies are currently underway to determine
whether increasing the sucrose concentration further can further increase the growth
rate.
While the model suggests that lengthening the duration of the misting cycle increases
the amount of nutrients delivered to the roots and should thereby increase growth, this
solution is actually more complex. For reasons as yet unknown, the misting cycle plays
a significant role in the successful operation of a mist bioreactor. Liu et al. [69] found
that a misting cycle of 3 min on / 30 min off was the optimum of those tested for
transformed roots of A. annua grown in their nutrient mist bioreactor, though its design
and operating conditions were different than those implemented by Weathers et al. [5].
Liu et al. [69] provided gas either only when mist was not being generated, or
129
www.taq.ir
continuously; while Weathers et al. [5] provided gas only when the mist was provided.
Interestingly, DiIorio et al. [60] also observed that hairy roots seemed to have optimum
mist duration for growth. Their studies with hairy roots of Beta vulgaris and Carthamus
tinctorius showed that either increasing or decreasing the “off” time beyond a certain
limit adversely affected root growth of those species. Chatterjee et al. [9] found that a
mist cycle of 1 min on / 15 min off caused transformed roots of A. annua to darken and
become necrotic after 12 d. Yet, studies with a single transformed root of A. annua [61]
showed that a mist cycle of 1 min on / 15 min off promoted healthier-looking roots and
higher fresh final biomass yields versus the other cycles tested. Studies by Towler
(unpublished results) in which the misting cycle was modified so that the mass flow rate
of sucrose was maintained while the sucrose concentration varied indicated that root
growth could be increased by increasing the length of misting cycle while decreasing
the mist off time. These results support the hypothesis that in a mist reactor, higher
growth yields can be achieved with increased droplet deposition and by manipulating
the on/off cycle period.
Droplet size and orientation of flow must also be considered for optimal growth and
secondary metabolite production of hairy root cultures. If the droplet size is too large,
the formation of a liquid layer along the root surface will impede gas transfer to the
roots and the system will behave as if it were a liquid-phase reactor [62]. Similarly,
when mist is provided in an upward direction, the mist can coalescence on the roots
closest to the mist feed with less mist reaching the tissue in the higher layers of the
growth chamber. Liu et al. [69] constructed an upward-fed mist reactor with three layers
of stainless steel mesh to support the roots, and found that there was a greater than 50%
decrease in biomass between the first (bottom) layer and the second and third layers.
Likewise, necrosis was observed in hairy roots of clone YUT16 of A. annua using
upflow mist delivery [9], but not with downflow mist delivery [61]. It is also likely that
as the root bed becomes very dense, the lower sections will accumulate liquid and
essentially become submerged. Mist reactors that are top fed have the advantage of cocurrent down-flow of gas and liquid phases along with gravity, which facilitates
drainage. In contrast, top versus bottom mist feeding seems to be of less consequence in
micropropagation systems and the orientation chosen is often a matter of convenience.
Another factor that may play a role in the growth of hairy roots is that of conditioned
medium. Both Chatterjee et al. [9] and Wyslouzil et al. [61] used autoclaved medium
with varying degrees of pre-conditioning by pre-growing roots in the medium before
using it in subsequent experiments. Wyslouzil et al. [61] also showed that there were
higher branching rates when roots were grown in conditioned medium versus fresh
medium. The identity of these “conditioning factors” remains elusive, although studies
have characterized some of them as oligosaccharides [70], peptides [71], and auxins
[64]. For consistency, it is recommended that fresh, filter-sterilized medium be used in
all experiments [72]. Work from our lab has routinely used filter-sterilized medium for
experiments since 1999.
6. Conclusions
Plant tissues are highly responsive to gases in their environment, especially O2, CO2,
and ethylene. Due to the low solubility of these gases, mass transfer of these gases to
130
www.taq.ir
the roots is hindered in a liquid system. Attempting to enhance gas transport by stirring,
bubbling, or sparging the liquid can damage shear-sensitive plant tissues. Therefore,
gas-phase reactors show many advantages over liquid-phase reactors, especially in
terms of the ability to easily manipulate gas composition in micropropagation chambers
and allow effective gas exchange in densely growing biomass. However, the
interactions between plant tissues and the nutrient mist environment can be complex
with many differing design aspects dictated by the application. For example, compare
the design of the growth chamber and the misting regimens required for growing hairy
roots vs. micropropagated plantlets (Figures 1 and 2). A better understanding of the
biological responses of the cultured tissues must be developed in order for mist reactors
to be exploited to their fullest potential. Recent results are promising and further studies
are warranted.
Acknowledgements
The authors thank Sev Ritchie for assistance with reactor construction, and the
following agencies for funding some of the described work: DOE P200A50010-95,
NSF BES-9414858, USDA 93-38420-8804, and NIH 1R15 GM069562-01.
References
[1] Weathers, P.J. and Zobel, R.W. (1992) Aeroponics for the cultures of organisms, tissues and cells.
Biotech. Adv. 10: 93-115.
[2] Friberg, J.A.; Weathers, P.J. and Gibson, D.G. (1992) Culture of amebocytes in a nutrient mist bioreactor.
In Vitro Cell. Dev. Biol.-Plant 28A: 215-217.
[3] Kim, Y.; Wyslouzil, B.E. and Weathers, P.J. (2002) Secondary metabolism of hairy root cultures in
bioreactors. In Vitro Cell. Dev. Biol.-Plant 38: 1-10.
[4] Perry, R.H. and Green, D.W. (1997) Perry's Chemical Engineer's Handbook, 7th ed., McGraw-Hill, New
York; pp.14-82.
[5] Weathers, P.J.; Wyslouzil, B.E.; Wobbe, K.K.; Kim, Y.J. and Yigit, E. (1999) Workshop on bioreactor
technology. The biological response of hairy roots to O2 levels in bioreactors. In Vitro Cell. Dev. Biol.Plant 35: 286-289.
[6] Tisserat, B.; Jones, D. and Galletta, P.D. (1993) Construction and use of an inexpensive in vitro ultrasonic
misting system. Hort. Technol. 3: 75-78.
[7] Woo, S.H. and Park, J.M. (1993) Multiple shoot culture of Dianthus caryophyllus using mist culture
system. Biotechnol. Techn. 7: 697-702.
[8] Buer, C.S.; Correll, M.J.; Smith, T.C.; Towler, M.J.; Weathers, P.J.; Nadler, M.; Seaman, J. and Walcerz,
D. (1996) Development of a nontoxic acoustic window nutrient-mist bioreactor and relevant growth data.
In Vitro Cell. Dev. Biol.-Plant 32: 299-304.
[9] Chatterjee, C.; Correll, M.J.; Weathers, P.J.; Wylslouzil, B.E. and Walcerz, D.B. (1997) Simplified
acoustic window mist bioreactor. Biotechnol. Techn. 11: 155-158.
[10] Correll, M.J.; Wu, Y. and Weathers, P.J. (2001) Controlling hyperhydration of carnations (Dianthus
caryophyllus L.) grown in a mist reactor. Biotechnol. Bioeng. 71: 307-314.
[11] Correll, M.J. and Weathers, P.J. (2001) One-step acclimatization of plantlets using a mist reactor.
Biotechnol. Bioeng. 73: 253-258.
[12] Correll, M.J. and Weathers, P.J. (2001) Effects of light, CO2 and humidity on carnation growth,
hyperhydration and cuticular wax development in a mist reactor. In Vitro Cell. Dev. Biol.- Plant 37: 405413.
[13] Bais, H.P.; Suresh, B.; Raghavarao, K.S.M.S. and Ravishankar, G.A. (2002) Performance of hairy root
cultures of Cichorium intybus L. in bioreactors of different configurations. In Vitro Cell. Dev. Biol.Plant 38: 573-580.
131
www.taq.ir
[14] Ilan, A. and Khayat, E. (1997) An overview of commercial and technological limitations to marketing of
micropropagated plants. Acta Hort. 447: 642-648.
[15] DeBergh, P.C. and Maene L.J. (1981) A scheme for commercial micropropagation of ornamental plants
by tissue culture. Sci. Hort. 14: 336-345.
[16] Cassells, A.C. (1997) Pathogen and microbial contamination management in micropropagation - an
overview. In: Cassels, A.C. (Ed.) Pathogen and Microbial Contamination Management in
Micropropagation. Kluwer Academic Publishers, Dordrecht, The Netherlands; pp. 1-14.
[17] Sutter, E.G.; Shackel, K. and Diaz, J.C. (1992) Acclimatization of tissue cultured plants. Acta Hort. 314:
115-119.
[18] Hempel, M. (1993) From micropropagation to "microponics". Practical Hydroponics International,
November/December: 21-23.
[19] Zobel, R.W. (1987) Gaseous compounds of soybean tissue culture: carbon dioxide and ethylene
evolution. Environ. Exp. Bot. 27: 223-226.
[20] Lee, C.W.T. and Shuler, M.L. (1991) Different shake flask closures alter gas phase composition and
ajmalicine production in Catharanthus roseus cell suspensions. Biotechnol. Techn. 5: 173-178.
[21] Kozai, T. (1991) Photoautotrophic micropropagation. In Vitro Cell. Dev. Biol.- Plant 27: 47-51.
[22] Ziv, M. (1991) Vitrification: morphological and physiological disorders of in vitro plants. In: DeBerge,
P.C. and Zimmerman, R.H. (Eds.) Micropropagation. Kluwer Academic Publishers, Dordrecht, The
[23] Kirdmanee, C.; Kitaya, Y. and Kozai, T. (1995) Effects of CO2 enrichment and supporting material in
vitro on photoautotrophic growth of Eucalyptus plantlets in vitro and ex vitro. In Vitro Cell. Dev. Biol.Plant 31: 144-149.
[24] Diaz-Perez J.C.; Shackel K.A. and Sutter E.G. (1995) Effects of in vitro-formed roots and
acclimatization on water status and gas exchange of tissue cultured apple shoots. J. Am. Soc. Hort. Sci.
120: 435-440.
[25] Lowe, K.C.; Anthony, P.; Power, J.B. and Davey, M.R. (2003) Invited review: novel approaches for
regulating gas supply to plant systems in vitro: application and benefits of artificial gas carriers. In Vitro
Cell. Dev. Biol.- Plant 39: 557-566.
[26] Ziv, M. (2000) Bioreactor technology for plant micropropagation. In: Janick, J. (Ed.) Horticultural
Reviews. John Wiley and Sons, New York; pp.1-30.
[27] Nairn, B.J.; Furneax, R.H. and Stevenson, T.T. (1995) Identification of an agar constituent responsible
for hydric control in micropropagation of radiata pine. Plant Cell Tissue Org. Cult. 43: 1-11.
[28] Kanechi, M.; Ochi, M.; Abe, M.; Inagaki, N. and Mackawa, S. (1998) The effects of carbon dioxide
enrichment, natural ventilation, and light intensity on growth, photosynthesis, and transpiration of
cauliflower plantlets cultured in vitro photoautotrophically and photomixotrophically. J. Am. Soc. Hort.
Sci. 123: 176-181.
[29] Solarova, J. and Pospisilova, J. (1997) Effects of carbon dioxide enrichment during in vitro cultivation
and acclimation to ex vitro conditions. Biol. Plant. 39: 23-30.
[30] Pospisilova, J.; Ticha, I.; Kadlecek, P.; Haisel, D. and Plzakova, S. (1999) Acclimatization of
micropropagated plants to ex vitro conditions. Biol. Plant. 42: 481-497.
[31] Fila, G.; Ghashghaie, J.; Hoarau, J. and Cornic, G. (1998) Photosynthesis, leaf conductance, and water
relations of in vitro cultured grapevine rootstock in relation to acclimatization. Physiol. Plant. 102: 411418.
[32] Kurata, K.; Ibaraki, Y. and Goto, E. (1991) System for micropropagation by nutrient mist supply. Am.
Soc. Agricult. Engineers 34: 621-624.
[33] Liu, C.Z.; Guo, C.; Wang, Y.C. and Ouyang, F. (2002) Comparison of various bioreactors on growth and
artemisinin biosynthesis of Artemisia annua L. shoot cultures. Process Biochem. 39: 45-49.
[34] Weathers, P.J. and Giles, K.L. (1988) Regeneration of plants using nutrient mist culture. In Vitro Cell.
Dev. Biol.- Plant. 24: 727-732.
[35] Cheetham, R.D.; Weathers, P.; DiIorio, A.; Glubiak, M. and Mikloiche, C. (1990) In vitro growth of a
recalcitrant male asparagus cultivar. Abstracts VII Intl. Congress on Plant Tissue and Cell Culture.
Amsterdam, The Netherlands, 24-29 June, 94.
[36] Cheetham, R.D.; Mikloiche, C.; Glubiak, M. and Weathers, P. (1992) Micropropagation of a recalcitrant
male asparagus clone (MD 22-8). Plant Cell Tissue Org. Cult. 31: 15-19.
[37] Weathers, P.J.; Cheetham, R.D. and Giles, K.L. (1988) Dramatic increases in shoot number and lengths
for Musa, Cordyline, and Nephrolepis using nutrient mists. Acta Hort. 230: 39-44.
132
www.taq.ir
[38] Hao, Z.; Ouyang, F.; Geng, Y.; Deng, X.; Hu, Z. and Chen, Z. (1998) Propagation of potato tubers in a
nutrient mist bioreactor. Biotechnol. Techn. 12: 641-644.
[39] Mavituna, F. and Park, J.M. (1986) Improvements relating to biotransformation reactions. International
Patent Application # PCT/GB85/00508.
[40] Honda, H.; Liu, C. and Kobayashi, T. (2001) Large-scale plant micropropagation. Adv. Biochem. Eng.
Biotechnol. 72: 157-182.
[41] Balandrin, M.F.; Klocke, J.A.; Wurtele, E.S. and Bollinger, W.H. (1985) Natural plant chemicals:
sources of industrial and medicinal materials. Science 228: 1154-1160.
[42] Banerjee, S.; Rahman, L.; Uniyal, G.C. and Ahuja, P.S. (1998) Enhanced production of valepotriates by
Agrobacterium rhizogenes induced hairy root cultures of Valeriana wallichii DC. Plant Sci. 131: 203208.
[43] Kittipongpatana, N.; Hock, R.S. and Porter, J.R. (1998) Production of solasodine by hairy root, callus,
and cell suspension cultures of Solanum aviculare. Forst. Plant Cell Tissue Org. Cult. 52: 133-143.
[44] Flores, H.E. and Curtis, W.R. (1992) Approaches to understanding and manipulating the biosynthetic
potential of plant roots. Ann. NY Acad. Sci. 665: 188-209.
[45] Kim, Y.; Wyslouzil, B.E. and Weathers, P.J. (2001) A comparative study of mist and bubble column
reactors in the in vitro production of artemisinin. Plant Cell Rep. 20: 451-455.
[46] Souret, F.F.; Kim, Y.J.; Wyslouzil, B.E.; Wobbe, K.K. and Weathers, P.J. (2003) Scale-up of Artemisia
annua L. hairy roots cultures produces complex patterns of terpenoid gene expression. Biotechnol.
Bioeng. 83: 653-667.
[47] Palazon, J.; Mallol, A.; Eibl, R.; Lettenbauer, C.; Cusido, R.M. and Pinol, M.T. (2003) Growth and
ginsenoside production in hairy root cultures of Panax ginseng using a novel bioreactor. Planta Medica
69: 344-349.
[48] Flores, H.E.; Yao-Rem, D.; Cuello, J.L.; Maldonado-Mendoza, I.E. and Loyola-Vargas, V.M. (1993)
Green roots: photosynthesis and photoautotrophy in an underground plant organ. Plant Physiol. 101:
363-371.
[49] Taya, M.; Sato, H.; Masahiro, K. and Tone, S. (1994) Characteristics of pak-bung green hairy roots
cultivated under light irradiation. J. Ferment. Bioeng. 78: 42-48.
[50] Mandoli, D.F. and Briggs, W.R. (1982) Optical properties of etiolated plant tissues. Proc. Natl. Acad.
Sci. 79: 2902-2906.
[51] Mandoli, D.F. and Briggs, W.R. (1983) Physiology and optics of plant tissues. What's New in Plant
Physiology 14: 13-16.
[52] McKelvey, S.A.; Gehrig, J.A.; Holar, K.A. and Curtis, W.R. (1993) Growth of plant root cultures in
liquid- and gas-dispersed reactor environments. Biotechnol. Prog. 9: 317-322.
[53] Curtis, W.R. (1993) Cultivation of roots in bioreactors. Curr. Opin. Biotechnol. 4: 205-210.
[54] Curtis, W.R. (2000) Hairy roots, bioreactor growth. In: Spier, R.E. (Ed.) Encyclopedia of Cell
Biotechnology. John Wiley and Sons, New York; pp. 827-841.
[55] Kim, Y.J.; Weathers, P.J. and Wyslouzil, B.E. (2002a) Growth of Artemisia annua hairy roots in liquidand gas-phase reactors. Biotechnol. Bioeng. 80: 454-464.
[56] Ramakrishnan, D.; Salim, J. and Curtis, W.R. (1994) Inoculation and tissue distribution in pilot-scale
plant root culture bioreactors. Biotechnol. Techn. 8: 639-644.
[57] Wilson, D.G. (1997) The pilot-scale cultivation of transformed roots. In: Doran, P.M. (Ed.) Hairy Roots:
culture and applications. Gordon and Breach / Harwood Academic, UK; pp. 179-190.
[58] Towler, M.J. and Weathers, P.J. (2003) Adhesion of plant roots to poly-L-lysine coated polypropylene
substrates. J. Biotechnol. 101:147-155.
[59] Kim, Y.J. (2001) Assessment of bioreactors for transformed root cultures. Ph.D thesis, Worcester
Polytechnic Institute, Worcester, MA.
[60] DiIorio, A.A.; Cheetham, R.D. and Weathers, P.J. (1992) Growth of transformed roots in a nutrient mist
bioreactor: reactor performance and evaluation. Appl. Microbiol. Biotechnol. 37: 457-462.
[61] Wyslouzil, B.E.; Waterbury, R.G. and Weathers, P.J. (2000) The growth of single roots of Artemisia
annua in nutrient mist bioreactors. Biotechnol. Bioeng. 70:143-150.
[62] Weathers, P.J. and Wyslouzil, B.E. (2000) Bioreactors, mist. In: Spier, R.E. (Ed.) Encyclopedia of Cell
Technology. John Wiley and Sons, New York; pp. 224-230.
[63] Biondi, S.; Lenzi, C.; Baraldi, R. and Bagni, N. (1997) Hormonal effects on growth and morphology of
normal and hairy roots of Hyoscyamus muticus. J. Plant Growth Regul.16: 159-167.
[64] Weathers, P.J.; Bunk, G. and McCoy, M. (2005) The effect of phytohormones on growth and artemisinin
production in Artemisia annua hairy roots. In Vitro Cell. Dev. Biol.- Plant, accepted for publication.
133
www.taq.ir
[65] Sung, L.S. and Huang, S.Y. (2000) Headspace ethylene accumulation in Stizolobium hassjoo hairy root
culture producing L-3,4-dihydroxyphenylalanine. Biotechnol. Lett. 22: 875-878.
[66] Wyslouzil, B.E.; Whipple, M.; Chatterjee, C.; Walcerz, D.B.; Weathers, P.J. and Hart, D.P. (1997) Mist
deposition onto hairy root cultures: aerosol modeling and experiments. Biotechnol. Prog. 13: 185-194.
[67] Crawford, M. (1976) Air Pollution Control Theory. McGraw-Hill, New York; pp.424-433.
[68] Friedlander, S.K. (1977) In: Smoke, Dust and Haze: Fundamentals of Aerosol Behavior. Wiley, New
York.
[69] Liu, C.Z.; Wang, Y.C.; Zhao, B.; Guo, C.; Ouyang, F.; Ye, H.C. and Li, G.F. (1999) Development of a
nutrient mist bioreactor for growth of hairy roots. In Vitro Cell. Dev. Biol.- Plant 35: 271-274.
[70] Schroder, R.; Gertner, F.; Steinbrenner, B.; Knoop, B. and Beiderbeck, R. (1989) Viability factors in
plant suspension cultures – some properties. J. Plant Physiol. 135: 422-427.
[71] Matsubayashi, Y. and Sakagami, Y. (1996) Phytosulfokine, sulfated peptides that induce the
proliferation of single mesophyll cells of Asparagus officinalis L. Proc. Natl. Acad. Sci. USA 93: 76237627.
[72] Weathers, P.J.; DeJesus-Gonzalez, L.; Kim, Y.J.; Souret, F.F. and Towler, M. (2004) Alteration of
biomass and artemisinin production in A. annua hairy roots by media sterilization method and sugars.
Plant Cell Rep. DOI: 10.1007/s00299-004-0837-4.
[73] Weathers, P.J.; DiIorio, A.A. and Cheetham, R.D. (1989) A bioreactor for differentiated plant tissues. In:
Proceedings of the Biotech USA Conference, San Francisco, CA, 247-256.
[74] Wilson, P.D.G.; Hilton, M.G.; Meehan, P.T.H.; Waspe, C.R. and Rhodes, M.J.C. (1990) The cultivation
of transformed roots from laboratory to pilot plant. In: Nijkamp, H.J.J.; van der Plas, L.H.W. and van
Aartrijk, J. (Eds.) Progress in Plant Cellular and Molecular Biology. Kluwer Academic Publishers,
[75] Nuutila, A.M.; Lindqvist, A.S. and Kauppinen, V. (1997) Growth of hairy root cultures of strawberry
(Fragaria x. ananassa Duch.) in three different types of bioreactors. Biotechnol. Techn. 11: 363-366.
[76] Whitney, P.J. (1990) Novel bio-reactors for plant root organ cultures. Abstracts VII Intl. Cong. Plant
Tissue Cell Cult., Amsterdam, The Netherlands; Abstract C4-19, 342.
[77] Whitney, P. (1992) Novel bioreactors for the growth of roots transformed by Agrobacterium rhizogenes.
Enz. Microbiol. Technol. 14: 13-17.
134
www.taq.ir
BIOREACTOR ENGINEERING FOR RECOMBINANT PROTEIN
PRODUCTION USING PLANT CELL SUSPENSION CULTURE
WEI WEN SU
Department of Molecular Biosciences and Bioengineering, University of
Hawaii, Honolulu, Hawaii 96822, USA – Fax: 1-808-956-3542 –
1. Introduction
Plant cell culture has long been considered as a potential system for large-scale
production of secondary metabolites. In recent years, with the advances in plant
molecular biology, plant cell culture has also attracted considerable interests as an
expression platform for large-scale production of high-value recombinant proteins.
Many plant species can now be genetically transformed. Callus cells derived from the
transgenic plants can be grown in simple, chemically defined liquid media to establish
transgenic cell suspension cultures for recombinant protein production. For certain plant
species, such as tobacco, it is also possible to establish transgenic suspension cell
cultures by directly transforming wild-type cultured cells. There are several notable
benefits of using plant suspension cultures for recombinant protein production. Plant
cells, unlike prokaryotic hosts, are capable of performing complex post-translational
processing, such as propeptide processing, signal peptide cleavage, protein folding,
disulfide bond formation and glycosylation, which are required for active biological
functions of the expressed heterologous proteins [1]. Plant cells are also easier and less
expensive to cultivate in liquid media than their mammalian or insect cell counterparts.
The potential human pathogen contamination problem associated with mammalian cell
culture does not exist in plant cell culture since simple, chemically defined media are
used [2]. When compared with transgenic plants, cultured plant cells also possess a
number of advantages. Cultured plant cells have a much shorter growth cycle than that
of transgenic plants grown in the field. Plant cell cultures are grown in a confined
environment (i.e. enclosed bioreactor) and hence devoid the GMO release problem.
Furthermore, cell suspension cultures consist of dedifferentiated callus cells lacking
fully functional plasmodesmata and hence there is minimum cell-to-cell communication.
This may reduce systemic post-transcriptional gene silencing (PTGS) which is believed
to be transmitted via plasmodesmata and the vascular system [3,4]. On the down side,
plant cells generally have a longer doubling time than bacterial or yeast cells. Genetic
instability associated with de-differentiated callus cells due to somaclonal variation is
another potential drawback in using cultured plant cells for recombinant protein
production. Due in part to their more evolved and more tightly controlled gene/protein
135
© 2008 Springer.
www.taq.ir
W.W. Su
regulation machinery, it is more difficult to manipulate protein expression in plant cells,
rending a generally lower protein expression level, normally between 0.1-1 mg L-1 of
culture [2], although product level as high as 129 mg L-1 has also been reported in the
case of recombinant human granulocyte-macrophage colony stimulating factor (hGMCSF) production in transgenic rice cell suspension culture [5].
Plant cell cultures have been used for producing a variety of recombinant proteins.
Several research groups have reported expression of antibodies or antibody fragments in
plant cell suspension cultures. Some notable examples are the expression of a secretory
anti-phytochrome single-chain Fv (scFv) antibody [6], a TMV-specific recombinant
full-size antibody [7], a mouse IgG1 recognizing a cell-surface protein of Streptococcus
mutants [8], and a mouse scFv [7,9], all using tobacco suspension culture. A number of
therapeutic proteins have also been expressed in plant cell cultures, including Hepatitis
B surface antigen (HBsAg) [10], human cytokines such as Interleukin IL-2, IL-4 [11],
IL-12 [12], and GM-CSF [5,13], ribosome-inactivating protein [14], and human D1antitrypsin [15,16]. Readers are also referred to other comprehensive reviews on the
subject of recombinant protein expression in plant tissue cultures [2,4,17].
Plant cell culture processes for recombinant protein production resemble
conventional recombinant fermentation processes in that they also encompass upstream
and downstream processing. However, there are distinctive properties associated with
plant cells that call for unique approaches in designing and operating plant cell
bioprocesses. The emphasis of this review will be on the upstream processing;
specifically, on the engineering considerations associated with the design and operation
of bioreactors for recombinant protein production using plant cell suspension cultures.
While much of the knowledge derived from the development of plant cell bioreactors for
secondary metabolite production are still relevant, issues unique to recombinant protein
production will be emphasized in this chapter. New findings since the publications of
other recent reviews of plant cell bioreactor [18,19] will be highlighted. Effective
bioreactor design and operation assures high productivity which is key to successful
bioprocess development. This chapter will begin with an overview of the unique
properties of plant cell cultures relevant to bioreactor design. Next, characteristics of
recombinant protein expression in plant cell culture are reviewed. This is followed by
discussions on a number of key topics relevant to bioreactor engineering, including
plant cell bioreactor operating strategies, bioreactor configurations and impeller design,
and innovative process sensing, as pertinent to recombinant protein production.
2. Culture characteristics
Plant cell suspension cultures are derived from callus cells. These are unorganized,
generally undifferentiated cells [20]. When suspended in liquid media, cells are
sloughed off friable calli to form a culture suspension. An effective plant cell
suspension culture system for recombinant protein production is expected to possess
certain desirable features, including fast growth rate, ease of genetic transformation,
high protein expression capacity, low endogenous proteolytic activity, low content of
phenolics (which may form complexes with proteins and complicates protein
purification) and other phytochemicals (such as oxalic acid) that may interfere with
downstream processing, superior post-translational processing capability, and good
136
www.taq.ir
Bioreactor engineering for recombinant protein production using plant cell suspension culture
culture stability (i.e. with low degrees of somaclonal variation and transgene silencing).
The most widely reported host species for developing plant suspension cultures to
produce recombinant proteins is tobacco (Nicotiana tabacum), followed by rice (Oryza
sativa). Other plant species such as tomato [21] and ginseng [22] have also been used.
Tobacco suspension culture is most widely used owing to its desirable growth
characteristics and ease of genetic transformation. However, it has been reported that
recombinant hGM-CSF is subject to more severe proteolytic degradation in the tobacco
cell culture medium than in the rice culture medium [5]. Therefore, while tobacco is a
convenient host, plant host species remains a factor to be considered in optimizing
recombinant protein production in plant suspension cultures. As far as bioreactor
development is concerned, the most relevant culture characteristics for recombinantprotein production include:
x Cell morphology, degree of aggregation, and culture rheology
x Foaming and wall growth
x Shear sensitivity
x Growth rate, oxygen demand, and metabolic heat evolution.
2.1. CELL MORPHOLOGY, DEGREE OF AGGREGATION, AND CULTURE
RHEOLOGY
Plant cells in suspension cultures display a range of shapes, with the largely spherical
and the rod (sausage-like) shapes being the most common. Size of single plant cells is
typically in the range of 50-100 Pm. Suspension cultures normally exhibit various
degrees of cell aggregation with the aggregate sizes varying dependent on the plant
species, growth stage, and culture conditions. While some plant cells form fine
suspensions with few large aggregates (with the largest aggregates smaller than 1 mm),
such as N. tabacum [5,23] and Anchusa officinalis [24], others form huge aggregates as
large as 2 cm in diameter, as in the case of Panax ginseng suspension culture used in the
Nitto Denko ginseng process [25] (cited in [18]). Formation of cell aggregates is mainly
due to the tendency of the cells to not separate after division. Cell adhesion due to the
presence of cell wall extra-cellular polysaccharides may enhance cell clumping
especially in the later stages of growth [26] (and references cited within). Cell
aggregates may consist of a mixture of highly mitotic and less mitotic cells (the latter
usually have greater potential for cellular differentiation into adventitious tissues or
organs). Cell aggregation promotes cellular organization and differentiation which is
generally believed to benefit secondary metabolite production, although in some cases
secondary metabolite production was found to be independent of aggregate sizes, such
as ajmalicine production in Catharanthus roseus culture [27] (cited in [18]). It appears
what is important for secondary metabolite production may not be the size of the
aggregates, but the state of organization and cellular differentiation within the cell
aggregates, which may not be entirely dependent on the aggregate size. For recombinant
protein production, cellular organization and differentiation potential is not as important,
and thus cell aggregation is generally viewed as undesirable since such feature
complicates the bioreactor operation. To this end, presence of oxygen/nutrient gradients
in complex cell clumps and sedimentation of large cell clumps are two apparent
problems. Formation of large cell clumps also complicates fluid pumping of the culture
137
www.taq.ir
W.W. Su
broth for downstream processing. Separating and dispersing the cells from the
aggregates by mechanical means in a bioreactor (e.g. by increasing bioreactor agitation)
is usually not very effective and may lead to cell damage, or even larger aggregates
[18]. Addition of pectinase and cellulase, higher cytokinin concentration, or lower
calcium concentration in the media may help to reduce the aggregate size [28].
However, the high cost associated with adding the hydrolytic enzymes at large scale
prevents the use of such strategy in industrial bioprocesses. It has been shown that overexpression of bacterial secretory cellulases leads to improved plant biomass conversion
[29]. It is plausible, therefore, to engineer plant cells to over-express cell wall bound or
secreted pectinase and/or cellulase as a means to control aggregate size in the
suspension culture; although its feasibility is yet to be tested.
Similar to the degree of cell aggregation, cultured cell morphology also depends on
the plant species, growth stage, and culture conditions. Suspension tobacco cell cultures
are often used for the expression of recombinant proteins. Under usual batch culture
conditions (e.g. in commonly used MS or B5 medium supplemented with auxin 2,4 -D
and 2-3% sucrose or glucose), the majority of suspension tobacco cells typically form
un-branched chains consisting of multiple sausage-shaped cells. Plant cell elongation
occurs after cell division ceases [30], it is tightly regulated (e.g. controlled by expansin
[30] and is believed to involve polar auxin transport [31]. Arrest of cell cycle by overexpressing cell-cycle inhibitor, while stops cell division, may also lead to cell
elongation [32]. Curtis and Emery [33] reported that when carrot cultures maintained on
a 7-day subculture interval were switched to a 14-day subculture interval, the cells
changed from spherical to elongated morphology. It is plausible, in the culture with a
longer subculture interval, cell division was slowed down due to nutrient limitation and
cell elongation was switched on. Cell elongation characteristics thus might be altered by
adjusting the nutrient regime and/or the types and concentrations of auxin (e.g. NAA is
known to promote cell elongation [31]) or by genetic manipulations (e.g. by altering the
expansin expression or by arresting the cell cycle). Note that elongated, filamentous
cells tend to entangle together to form a cellular network, resulting in higher packed cell
volume (PCV) for a given number of cells per reactor volume (than spherical cells), and
hence higher apparent viscosity. Curtis and Emery [33] reported the highly viscous and
power-law type rheological properties associated with batch-cultured tobacco
suspension cells were resulting from elongated cell morphology. The bioprocess
implication is significant in that less biomass can be attained with cultures of elongated
cells as opposed to spherical-shaped cells. When cultured in similar high-density
perfusion bioreactors, and under comparable growth conditions, tobacco cell culture
reached only 10 g/L dry weight (with PCV exceeding 60%), whereas A. officinalis cell
culture (which consists of mostly spherical cells and forms fine suspension with few
large aggregates) can reach cell dry weight over 35 g/L with PCV over 60% [34]. It may
be possible to use molecular approaches to reduce/block auxin efflux or to manipulate
cellulose biosynthesis (and hence cell wall composition and structure) to alter the
morphology of the cells.
Culture rheological property significantly affects bioreactor mixing, oxygen, and
heat transfer. It also affects how high cell concentrations can reach. In addition to cell
size, morphology and degree of aggregation, rheological property of suspension plant
cell culture is affected by cell concentration (especially in terms of biotic phase volume,
138
www.taq.ir
as opposed to cell numbers or cell dry weight) and cellular water content. Plant cell
suspension cultures are usually considered highly viscous. This view comes from the
fact that typically plant cell cultures reach a very high culture biotic phase volume
fraction (PCV over 50%) even in batch cultures. The culture spent media however
usually is not viscous and behaves as Newtonian fluid. Power-law models including
Bingham plastics, pseudoplastics, and Casson fluids have been applied to describe the
rheological properties of high-density plant cell suspension cultures [28,35]. In powerlaw rheological models,
W
Wo KJ n
where W is shear stress,
(1)
J is shear rate, K is consistency index, Wo is yield stress, and n is
the flow behaviour index. For pseudoplastic fluids, n < 1 and Wo = 0; for Bingham
plastics, n = 1, and Wo z 0. As stated earlier, cell morphology can have a considerable
influence on the culture rheological characteristics. Cultures consist of mainly large
non-friable cell aggregates form very heterogeneous particulate suspensions. At low cell
concentrations, these cultures typically behave more like a Newtonian fluid [33]. At
high cell concentrations, the presence of a large number of large, discrete cell
aggregates renders an unambiguous determination of the culture rheological properties
more difficult [28]. Cultures that consist of mainly large aggregates are generally shown
to be less viscous than those consists of elongated cells entangled into a filamentous
cellular network [33]. Most viscous high-density plant suspension cultures exhibit
shear-thinning, pseudoplastics characteristics [35,36]. In this case, the apparent culture
viscosity (Pa) is related to the shear rate as:
Pa
KJ n1
(2)
implying that apparent viscosity is lower under higher shear. As such, mixing and
bubble dispersion should be more efficient in the impeller region where high shear
exists, whereas the region away from the impeller may experience a higher apparent
viscosity and may lead to poor mixing and mass transfer. Another unique phenomenon
was noted recently during high density cultures of tobacco cells (PCV over 60%; Su,
unpublished) in a 3-L stirred-tank bioreactor, where cells became immobilized on
standard six-blade Rushton disc turbine impellers (i.e. impeller became completely
covered by a thick layer of plant cell biomass) to an extent that the impeller became
shaping like an elliptical object. Mixing and mass transfer efficiency dropped as a result.
This perhaps was triggered by an initial accumulation of entangled filamentous tobacco
cell clumps in the gas-filed cavities behind the impeller blades. Since this phenomenon
can cause considerable reduction in impeller performance, it warrants further
investigation. In some culture systems, yield stress has been reported (i.e. behaving as
Bingham fluids). The existence of a yield stress may impact aeration efficiency in a way
that small bubbles may experience a much longer residence time and become depleted
in oxygen [36]. Therefore, oxygen transfer in the bioreactor may not be efficient despite
a high gas hold-up. Manipulating culture medium osmotic pressure has been shown to
139
www.taq.ir
W.W. Su
reduce apparent culture viscosity in some studies [28,37]. However, increasing medium
osmotic pressure generally causes plasmolysis (shrinkage of cytoplasm within the cell)
but may not significantly reduce the overall cell size due to the presence of the rigid cell
wall. As such, its effect on reducing culture viscosity may not result from reducing the
cell size.
2.2. FOAMING AND WALL GROWTH
Plant cells are commonly cultured in bioreactors with bubble aeration, which produces
foaming at the culture broth surface. A number of factors are believed to attribute to
foam formation. These include presence of extra-cellular polysaccharides, proteinacious
substances, fatty acids (secreted or released by lysed cells), and high sugar concentration
during the early stage of cultivation [28,37]. Extent of foaming is affected by aeration
rate, medium composition, culture viscosity, biomass level, and the bioreactor
configuration [38]. As summarized in Abdullah et al. [37], common measurement
techniques and parameters for characterizing culture foaming include the ratio of foam
volume to gas flow rate, volume of liquid held in the foam, volumetric rate of foam
overflow, and decrease of foam volume with time. As a result of culture foaming, a
large amount of cells become entrapped in the foam layer, rendering reduced volumetric
biomass concentration in the broth. These foam-entrapped cells develop a thick crust
adhering to the reactor vessel and the probes. The accumulated cell crusts may become
necrotic and secrete inhibitory substances such as proteases or superannuated cell
organelles. Under severe foaming, foam overflow can clog the air vent filter and make
the culture susceptible to contamination. Wall growth is also know to affect the scale up
and dynamic operating characteristics of chemostats and bioreactor cultures with
substrate inhibition [39]. Several strategies have been employed to combat the
foaming/wall growth problem, including addition of silicone-based and polypropylene
glycol antifoam agents, mechanical foam disruption, reduced bubble aeration rate,
intermittent bubble aeration, bubble-free surface or membrane aeration, reduced calcium
concentration in the medium, coating of reactor vessel wall with Teflon or silicone, and
use of mechanical/magnetic scrapper units to push the wall-growth cell crusts back into
the culture broth. Since plant cells entrapped in the foam layer have the tendency to stick
to the antifoam sensor, it is difficult to use the conductive-type sensor commonly applied
in microbial fermentors for monitoring foam formation and effectively control the foam
by accurately triggering the automatic dosing of antifoam agents. Once the meringue-like
cell crust layer is built up above the broth, addition of antifoam agents becomes less
effective in suppressing further wall-growth development. Furthermore, overdosing of
antifoam agents may reduce oxygen transfer since antifoam reduces surface tension,
lowering the mobility of the gas/liquid interface and causing interfacial breakage [40].
Abdullah et al. [37] examined various strategies to overcome foaming and wall growth
in the culture of Morinda elliptica cell suspension culture and concluded that bubblefree aeration using thin-walled silicone membrane tubing was the only strategy capable
of completely eliminating wall-growth. Bubble-free membrane aeration however is not
suited for large-scale bioreactors due to reduced membrane surface to volume ratio and
hence reduced oxygen transfer upon scale-up. We found that at least in smaller benchscale bioreactors, silicone-based antifoam addition and use of a magnetic scrapper
(consists of two small but strong magnets, one placed on the interior reactor wall and
140
www.taq.ir
the other on the exterior wall to form a magnetic pair) can at least reduce the extent of
wall growth of transgenic tobacco cells cultured in a sparged stirred-tank bioreactor.
Under these circumstances, however, a significant foam layer still built up around the
rotating shaft and the probes, leads to biomass loss. We found that by using an impeller
installed above the culture broth to serve as a mechanical foam breaker was not
effective in breaking up foams. On the contrary, since the rotating speed of the impeller
is not sufficiently high, the cells entrapped in the foam layer actually formed a think
crust on the foam-breaker impeller. As mentioned earlier, this phenomenon was also
noted even for the impellers that were immersed in the culture broth. Since none of the
aforementioned strategies offer a practical solution to effectively eliminate foaming and
wall growth, it remains a challenge to overcome such problem in plant cell bioreactor
design. Fortunately, as the reactor is geometrically scaled up, the reactor cross-section
per volume ratio drops, and the wall growth problem is expected to reduce.
2.3. SHEAR SENSITIVITY
Cultured plant cells embrace vacuoles up to 95% of cell volume and their primary cell
wall is made of parallel cellulose micro fibrils embedded in a polysaccharide matrix.
Therefore, plant cells are generally considered shear sensitive. However, shear
sensitivity varies greatly among plant species and may be affected by the culture age.
Over the past two decades several studies have been conducted to investigate how
cultured plant cells respond to various shear environments. Major studies published
prior to 1993 had been summarized in a review by Meijer et al. [41]. More recently,
Kieran et al. [42] conducted a comprehensive review of the same subject. A number of
studies in this topical area have been published by Erick Dunlop’s group [43-45] and by
Kieran and co workers [42,46]. Studies of the sensitivity of cultured cells to
hydrodynamic forces are complicated by the difficulties to establish a defined
hydrodynamic environment mimics that of the bioreactors. Shear studies have been
conducted under well-defined laminar or turbulent flow conditions using capillary, jet,
and Couette flows [42]. One common shortcoming in these studies is that the flow
conditions in these model systems are not entirely representative of the complex
turbulent flow conditions in typical bioreactors. For shear studies conducted directly in
bioreactors, however, it is necessary to correlate cellular shear responses to some
quantifiable bioreactor parameters, owing to the poorly defined hydrodynamic
environment in the bioreactors. To this end, a number of physiological parameters have
been used as indicator of cellular shear response; these include loss of viability,
membrane integrity, respiratory (mitochondrial) activity, release of intracellular
components, and morphological variations [41,42]. Cellular response to hydrodynamic
shear is affected by the intensity as well as the exposure duration of the cells to shear
stress. In this context, the cumulative energy dissipation has been suggested as a useful
basis for correlating data from shear studies involving a wide range of plant species,
hydrodynamic conditions, and physiological indicators [19,42,43]. The cumulative
energy dissipation imposed on the cells per unit reactor working volume (Ec) can be
calculated using the following equation [19,43]:
141
www.taq.ir
W.W. Su
Pg
Ec
PI
³V
R
dt
³
Po
(UN p N i3 Di5 )I
VR
dt
(3)
where P is power input, VR is the reactor working volume, I is the biotic phase volume
fraction in the culture, t is time, Pg is gassed power input, Po is ungassed power input, U
is broth density, Np is the impeller power number, Ni is the impeller speed, and Di is the
impeller diameter. Figure 1 (reproduced from reference [18]) shows various shear
response indices obtained under a variety of flow conditions, as a function of Ec in three
different cell cultures. Each shear response index appears to be associated with a
threshold level of cumulative energy dissipation, beyond which extensive reduction in
cellular activity is noted. For instance, membrane integrity of Morinda citrifolia cells
was severely damaged at a critical cumulative dissipated energy level exceeding 108 Jm- 3
(Figure 1, curve d). Doran [19] compared the performance of various impeller designs
for plant cell bioreactors using a threshold Ec level of 107 Jm-3. Cumulative energy
dissipation serves as a convenient index for estimating hydrodynamic shear damage.
However, as indicated by Doran [19] and by Kieran [18], the application of this index
also has its limitations. Effect of hydrodynamic shear on the plant cells in an
aerated/stirred-tank bioreactor does not result entirely from the impeller power input;
under the same impeller power input, shear damage on the cells is also anticipated to
vary depending on the impeller geometry. Note that Ec is a global (average)
hydrodynamic property, and hence it does not reflect how the energy dissipation rates
are distributed within the reactor. The highest specific rates of energy dissipation occur
near the impellers, and impellers having different sweep volumes and trailing vortex
structures are expected to inflict different local shear conditions in the vicinity of the
impellers [19,44]. Doran [19] and Sowana et al. [44] also pointed out that for impellers
that produce more rapid broth circulation, cells are transported to the high-shear
impeller region more frequently and hence more shear damage is expected. Another
point to consider is that under gassing conditions, the impeller power input is reduced,
and hence the cumulative energy dissipation is expected to decrease according to
equation (3). While shear damage resulting from the hydrodynamic forces associated
with bubble rupture is believed to be insignificant in plant cell cultures [18,43], there is
no evidence indicating shear damage is reduced with increasing bubble aeration rates at
a fixed stirrer speed. The suitability of Ec as a common basis to quantify the agitationbased shear forces under different bubble aeration rates apparently warrants further
investigations.
142
www.taq.ir
Figure 1. Cellular shear responses as a function of Ec for Daucus carota [43], Morinda
citrifolia [46], and Atropa belladonna [38]. Shear response indicators: (a) aggregate size,
(b) cell lysis, (c) mitochondrial activity, (d) – (f) membrane integrity, (g) protein release,
and (h) cake permeability/aggregate size. Reproduced from Kieran, P. M. (2001) [18], with
permission from Taylor and Francis.
The biological basis for cell response to hydrodynamic shear is not well understood. It
has been hypothesized that calcium ion flux, osmotic regulation, cell–cell
contact/aggregation, and stress protein expression might be the key processes involved
in perception and responses to hydrodynamic shear [47]. In recent years, more
experimental evidence has emerged indicating oxidative burst as a potentially important
step in the signal transduction cascade that triggers the plant defence mechanism in
response to hydrodynamic shear. Shortly after pathogenic attack, plant cells usually
produce and release active oxygen species (AOS) at the cell membrane surface; these
include the superoxide radicals, the hydroxyl radicals, and hydrogen peroxide [18,48].
This is known as the oxidative burst. Yahraus et al. [49] were among the first to present
evidence for mechanically induced oxidative bursts in plant suspension cultures.
Recently, Han and Yuan [48] investigated the oxidative bursts in suspension culture of
Taxus cuspidate induced by short-term laminar shear under Couette flow condition.
They found that NAD(P)H oxidase is the key enzyme responsible for oxidative bursts
under shear, and the superoxide radical burst may cause changes in the membrane
permeability, while hydrogen peroxide burst plays an important role in activating
phenylalanine ammonia lyase and phenolic accumulation. Han and Yuan [48] further
postulated that G-protein, calcium channel, and phospholipase C may be involved in the
143
www.taq.ir
W.W. Su
signal transduction pathway of oxidative bursts induced by hydrodynamic shear, as
depicted in the model shown in Figure 2.
NAD(P)H
Oxidase
Induction of
secondary
metabolism
Figure 2. Hypothetical model proposed by Han and Yuan [48] for oxidative burst in
cultured plant cells induced by hydrodynamic shear. (S) shear stress; (G) G-protein; (R)
shear signal receptor in the cell membrane; (IP3) inositol phosphates; (DG) diacylglycerol;
(PLC) phospholipase C. Adapted from Han, R. and Yuan, Y. (2004) [48], with permission
from the American Chemical Society.
According to such model, it might be possible to engineer plant cell lines that are less
susceptible to shear damage by disrupting the signal pathway that leads to oxidative
bursts. Alternatively, shear induced genes might be identified using DNA microarray
and/or proteomics tools to further elucidate the biological basis of shear sensitivity.
Thus far, two notable approaches have been demonstrated to improve plant cell
tolerance to shear damage. One involves the selection of shear-tolerant strains [35] and
the other the application of a non-ionic surfactant, Pluronic® F-68. Sowana et al. [45]
reported beneficial effect of Pluronic® F-68, which has been demonstrated as an efficient
protection agent of mammalian and insect cells from shear damage, on protecting
cultured plant cells from hydrodynamic damage, and suggested that the protection
144
www.taq.ir
mechanism is likely to result from cell membrane manipulation (perhaps by reduction
of plasma membrane fluidity, leading to an increase in cellular resistance to shear).
2.4. GROWTH RATE, OXYGEN DEMAND, AND METABOLIC HEAT LOADS
For recombinant protein production, plant species that generate fast-growing cell
cultures are often preferred. Top the list are tobacco and rice cell cultures. Tobacco BY2 cells are particularly appealing because of their remarkably fast growth rate, as well as
the ease for Agrobacterium-mediated transformation and cell cycle synchronization.
Doubling time as short as 11 hours has been reported for the tobacco BY-2 cells [50].
Koroleva et al. [51] recently demonstrated that the growth rate of BY -2 cells can be
transiently increased by expressing a putative G1 cyclin gene, Antma;CycD1;1, from
Antirrhinum majus; this cyclin gene is known to be expressed throughout the cell cycle
in the meristem and other actively proliferating cells. Expression of cycD2 was also
shown to increase tobacco plant growth [52]. Effect of over-expressing cycD genes in
tobacco cell cultures on cell proliferation and recombinant protein production is
currently being investigated in our laboratory. Tobacco cell cultures derived from other
tobacco varieties, e.g. Xanthi, do not grow as fast as the BY-2 cells, but still has a
relatively short doubling time about 1.5-2 days. Gao and Lee [53] reported a doubling
time of about one day for tobacco NT-1 cells (which is similar to the BY-2 cells)
expressing E-glucuronidase (GUS). For rice cell culture, Trexler et al. [16] reported
doubling time of 1.5 ~ 1.7 days for a transgenic rice cell culture expressing human D1antitrypsin. Terashima et al. [15], on the other hand, reported a very long doubling time
of 6-7 days in their transgenic rice cell cultures expressing human D1-antitrypsin.
Maximum specific oxygen uptake rate was 0.78 ~ 0.84 mmol O2/(gdw h) in the
transgenic rice cell culture reported by Trexler et al. [16]; 0.4 ~ 0.5 mmol O2/(gdw h)
for the transgenic tobacco NT-1 cells expressing GUS [53]. Kieran [18] reported that
specific oxygen consumption rate for plant cell cultures is generally of the order of 10-6
g O2/(gdw s) (i.e. 0.11 mmol O2/(gdw h)). Gao and Lee [53] observed improved cell
growth, increased oxygen consumption rate, and GUS production with higher oxygen
supply [53]. In general, if expression of the recombinant protein is driven by a
constitutive promoter, expression is usually growth associated and hence factors that
promote cell growth (such as improved oxygen supply) are expected to promote protein
expression. Unlike plasmid-based expression in bacterial cells that lead to huge amount
of over-expression, the metabolic burden resulting from foreign protein expression in
plant cells is generally not high enough to substantially impact the cell growth or
oxygen demand, except if the foreign gene product is toxic or able to interact with the
plant metabolism to cause altered growth characteristics.
In cell cultures there generally exists a critical dissolved oxygen level, below which
linear (in lieu of exponential) growth is seen as a result of oxygen limitation. The critical
dissolved oxygen level in plant cell cultures is typically at 15 ~ 20% air saturation [36].
Based on the specific oxygen consumption rate, one can estimate the metabolic heat
evolution since the heat of reaction for aerobic metabolism is approximately -460 J per
mmol of oxygen consumed [54]. As cited in reference [18], metabolic heat evolution
rate of 138 J/(g dw h) was reported by Hashimoto and Azechi [55] in a large-scale
(6,340 litres of working volume) tobacco chemostat culture with an average cell density
145
www.taq.ir
W.W. Su
of 17 g/L. Therefore an oxygen demand of about 0.3 mmol O2/(gdw h) is estimated,
which is in good agreement with that reported by Gao and Lee [53]. Assuming
comparable heat transfer characteristics between high-density plant cell culture and
viscous fungal fermentation, Kieran [18] suggests that efficient heat removal in plant
cell bioreactors can be easily achieved even with moderate mixing.
Tolerance to low-oxygen stress by cultured plant cells is expected to be species
dependent. While physiological responses of bioreactor-cultured plant cells/hairy roots
to extended hypoxic stress (at the molecular level) is not well documented, it is
generally believed that engineering plant cells for improved hypoxic stress tolerance is
desirable, or even necessary, to complement the bioreactor design to combat the oxygen
supply problem in large-scale plant cell bioreactor, especially for high-density cultures.
Two notable approaches have been taken to engineer cultured plant cells and/or hairy
roots for improved tolerance to hypoxic stress. In one approach, it involves overexpression of bacterial or plant haemoglobin genes [56,57]. In another approach, Doran
and co-workers [58] found that hairy roots over-expressing either Arabidopsis pyruvate
decarboxylase or alcohol dehydrogenase, the two major enzymes in the fermentation
pathway, showed improved growth over control roots under microaerobic conditions.
3. Characteristics of recombinant protein expression
In bioreactor design, it is useful to relate the pattern of product synthesis to cell growth.
The production occurs either predominantly during active cell growth (i.e. growth
associated) or after active cell growth is ceased (i.e. non-growth associated). In
recombinant protein production, the production pattern is strongly affected by the type
of promoter used. When a constitutive promoter, such as the widely popular cauliflower
mosaic virus (CaMV) 35S promoter, is used to drive the transgene expression, the
recombinant protein production is considered largely growth associated. Cells may
continue to produce the recombinant protein upon initial entering into the stationary
phase of the growth cycle, but this is usually accompanied with increased proteolytic
activities, and hence the recombinant protein level tends to descend during the stationary
phase when the 35S promoter is used. If an inducible promoter is used, generally the
transgene is induced after the culture reaches a high biomass concentration in the
late/post exponential growth phase [59]. In this case, recombinant protein production is
decoupled from the active cell growth. A number of inducible promoters have been used
for expressing recombinant proteins in plant suspension cultures. The rice D-amylase
(RAmy3D) promoter which is induced by sugar starvation was used in rice cell cultures
to express recombinant D1-antitrypsin [15,16] and recombinant hGM-CSF [5]; the
Arabidopsis thaliana heat-shock (HSP18.2) promoter [60], the tomato light inducible
rbcS promoter [61], the methyl jasmonate inducible potato cathepsin D inhibitor (CDI)
promoter [59], the glucocorticoid-inducible GVG promoter [62], the sweet potato
oxidative stress-inducible peroxidase (POD) promoter [63], and the abscisic acid,
tetracycline, and copper inducible promoters [64], have all been examined in tobacco
cell cultures for recombinant protein production. In order to optimize the efficiency of
an inducible gene expression system, it is necessary to examine the inducer
concentration and timing of inducer addition. Depending on the nature of the inducer,
repeated inducer feeding may be desirable, and hence optimization of inducer feeding
146
www.taq.ir
would be necessary. Published data in this area for plant cell culture is scarce. Suehara
et al. [59] investigated optimal induction strategies for the expression of a GUS reporter
driven by the CDI promoter. Since the addition of the inducer, methyl jasmonate, led to
metabolic by-products that reduced cell growth, Suehara et al. [59] had to replace the
spent media with fresh ones to remove the potential inhibitory substances, and to devise
an inducer feeding strategy by keeping the inducer concentration within a narrow range.
Trexler et al. [16] postulated that expression systems based on the rice D-amylase
promoter could be further improved by optimizing the timing of medium exchange
using suitable physiological indicators, and by exposing the culture to consecutive
growth and sugar-starvation phases. Atsuhiko Shinmyo’s group [50,65] isolated several
growth-phase dependent strong promoters from tobacco BY-2 cells, based on the
principle that genes with low copy number in the genome, but with abundant transcripts
are likely controlled by a strong promoter. Among these, promoter fragments of two
genes that encode putative alcohol dehydrogenase and pectin esterase, respectively,
were found to strongly express during the stationary phase. Strong promoters active in
the stationary phase are good candidates for driving recombinant protein production in
high-density stationary-phase cultures (e.g. in high-density perfusion cultures) or
immobilized cell cultures [50].
In addition to the knowledge on how protein production pattern is related to the cell
growth pattern, it is useful to know whether the protein products are secreted into the
media. Recombinant proteins might be targeted to the ER-Golgi secretion pathway
using a proper signal peptide. It is highly desirable to enable effective secretion of the
protein product to simplify downstream protein purification. The secretory pathway also
provides a better cellular environment for protein folding and assembly than the cytosol,
since the endoplasmic reticulum contains a large number of molecular chaperones and
is a relatively oxidizing environment with low proteolytic activities, rendering generally
higher accumulation of the recombinant proteins [66]. However, there are exceptions to
the rule, suggesting the overall protein yield may also be affected by the intrinsic
properties of each protein product. Furthermore, it should be cautious that the extracellular compartment is not loaded with proteolytic activities that can degrade the
proteins of interests. Shin et al. [5] observed higher proteolytic activities in the tobacco
cell culture than in the rice cell culture. Addition of stabilization agents such as gelatin,
polyvinyl pyrrolidone (PVP), and bovine serum albumin (BSA) have met with various
degrees of success among the proteins tested for stabilization [2]. Comparing to these
common protein stabilizing agents, the peptide antibiotic bacitracin may be more
effective towards stabilizing a broader range of proteins; although at high
concentrations (1 mg/ml) bacitracin becomes toxic to plant cells [67]. Another strategy
to stabilize secreted recombinant proteins in plant suspension cultures is via in-situ
adsorption. James et al. [68] coupled an immobilized protein G and a metal affinity
column to a culture flask to recover secreted heavy-chain mouse monoclonal antibody
and histidine-tagged hGM-CSF, respectively, by recirculating the culture filtrates
through these columns. Increased product yields for both proteins were noted, resulting
from reduced protein degradation.
A variety of molecular strategies exist for improvement of gene expression and
heterologous protein accumulation in plants and plant cells [69]. General points of
consideration include the use of appropriate promoters, enhancers, and leader sequences
147
www.taq.ir
W.W. Su
[70]; optimization of codon usage; control of transgene copy number; sub-cellular
targeting of gene products (e.g., by using an ER-targeting signal peptide or ER-retention
HDEL or KDEL signal); the position in the plant genome at which the genes are
integrated [71]; and the removal of mRNA-destabilizing sequences [72]. In some cases,
nuclear matrix attachment regions (MARs) have been found to increase transgene
expression [73]. Viral genes that suppress PTGS, such as the potyvirus hc protease
genes, can be used to prevent transgene PTGS [74]. As plants expressing these genes
may have greatly increased virus susceptibility, this approach may not be practical for
field plants but could work well in suspension cells. Additional ways to increase
expression levels include the use of different plant species, integration-independent
expression, and enhancing correct protein folding by co-expressing disulfide isomerases
or chaperone proteins [69].
4. Bioreactor design and operation
The culture and production characteristics described in the preceding sections provide
the basis for effective bioreactor design and operation to produce recombinant proteins
from transgenic plant cell suspension cultures. In addition, it is important to incorporate
cellular stoichiometry, mass and energy balances, reaction kinetics, heat and mass
transfer, hydrodynamics and mixing, shear, and process monitoring and control, in
bioreactor design for transgenic plant cell cultures. General discussions on the topic of
plant cell bioreactors can be found in a number of comprehensive reviews. Two of the
more recent ones are from Doran [19] and Kieran [18]. Here we will focus on plant cell
bioreactor operating strategies, bioreactor configurations and impeller design, and
innovative process sensing, as pertinent to recombinant protein production.
4.1. BIOREACTOR OPERATING STRATEGIES
While it is most common to culture plant cells in the single-stage batch mode, two-stage
batch [15], fed-batch [23,59], chemostat [75], and perfusion modes [34,76] have also
been explored for protein production from cultured plant cells [4]. As discussed in the
previous section, choice of bioreactor operation mode is largely governed by the pattern
of product formation and the way the product is translocated following its synthesis. For
growth-associated, intracellular protein products (e.g. when a constitutive promoter and
an ER-retention signal are used), protein productivity could be improved by increasing
the cell growth rate and prolonging the active cell-growth phase in a single-stage batch
or fed-batch bioreactor. To increase biomass output, chemostat cultures generate a
constant stream of biomass, from which intracellular recombinant proteins can be
harvested. However, it is difficult to run a true chemostat at high biomass concentration
with plant cell suspensions, due to cell aggregation, slow growth, surface adhesion and
high viscosity at high biomass densities [28]. Semi-continuous cultivation, in which a
portion of the cell suspension is periodically removed and then replenished with fresh
medium, can be applied as an alternative to chemostat cultures. Biomass (and
recombinant protein) productivities may be further improved using perfusion culture
with a bleed stream. A much higher cell density can often be obtained in perfusion
cultures compared to continuous and semi-continuous cultures, because cells are
148
www.taq.ir
retained within the reactor via a cell retention device. With a bleed stream, the perfusion
reactor can be operated under a (quasi-) steady state at a very high cell concentration. It
is well known that for a culture system that follows simple Monod kinetics, the
maximum biomass output rate in a perfusion reactor with a bleed stream is higher than
that in a chemostat by a factor of 1/\, where \ is the bleed ratio (the ratio between flow
rates of the bleed stream and the feed stream). In a perfusion reactor, the specific growth
rate can be manipulated by adjusting the bleed stream. Another advantage of perfusion
operation is that inhibitory by-products in the spent medium can be removed efficiently,
since very high perfusion rates can be used without cell washout.
For growth-associated, extra-cellular protein products, one also needs to consider
increasing cell growth rate, prolonging active cell growth, and raising biomass output,
as for the growth-associated intracellular products; but since the product is secreted into
the media, one may also consider coupling a suitable protein recovery unit (such as an
affinity column) to the reactor by re-circulating the culture spent media through the
recovery unit to harvest the product [68]. If operated at the perfusion mode, a high
perfusion rate should be used with the bleed stream adjusted to give a high specific
growth rate.
For non-growth associated, intracellular protein products (e.g. when an inducible
promoter or a stationary-phase specific promoter is used along with an ER-retention
signal), two-stage batch cultures should be advantageous. Two-stage culture can be
conducted in one physical bioreactor unit or in two separate reactors. For the latter, if
the culture cycle in the growth stage is shorter than the production stage, one growth
stage reactor may be used to feed multiple production stage reactors [77]. The
production-stage operation strongly depends on the gene induction system used. For
instance, when a stationary-phase specific promoter [65] is used, it would be desirable to
prolong the stationary phase (i.e. the production stage) to increase the protein
production yield. One apparent challenge would be to provide a suitable culture regime
and cellular microenvironment that enable essentially zero net growth while avoids or at
least minimizes culture degradation (e.g. programmed cell death or elevated proteolytic
activity). In plant cell cultures, the net cell growth usually ceases when the packed cell
volume approaches ca. 60-70%. Under such high biomass volume fraction, the cellular
mitotic index generally is very low and the culture usually is not able to sustain a high
metabolic activity for very long. This problem can be alleviated, in part, by perfusing the
culture with fresh media [34]. Further improvements are expected to derive from deeper
understanding of the cellular responses to the biotic stress caused by the extremely high
biomass volume fraction. Sugar-starvation inducible D-amylase promoter has been used
to express hGM-CSF at a level as high as 129 mg/L [5]. When a sugar-starvation
inducible promoter is used, it is necessary to remove sugar from the media to induce
transgene expression. In such case, one needs to conduct a media exchange into a sugarfree nutrient solution [15] or solution containing an alternative carbon source [78], or to
supplement macro- and micro-nutrients into the sugar-depleted media at the end of the
growth stage, in a single reactor unit, or to concentrate the cells from the growth-stage
reactor and inoculate the cells into a second reactor to induce the protein expression. If a
single reactor unit is employed, medium exchange can be achieved by filtration or
sedimentation. If a chemical inducer is used to induce transgene expression, the inducer
may be fed into the culture at late growth stages and repeated inducer feeding may
149
www.taq.ir
W.W. Su
prolong and increase transgene expression. However, optimization of inducer dosage
and feeding strategy is dependent on the nature of the inducer (considering its toxicity
and chemical stability) and how the inducer activates the promoter. In principle, twostage chemostats may also be considered. Here the first stage chemostat is used to
provide the cells for the second stage chemostat, which is manipulated to enhance
product synthesis. A low dilution rate should be used in the second-stage chemostat to
reduce the cell growth rate. This could be done by increasing the reactor volume of the
second stage chemostat. One major drawback of this operation is that the low dilution
rate also reduces the biomass output rate and hence decreases the intracellular
recombinant protein productivity.
For non-growth associated, extra-cellular protein products, it would be
advantageous to employ fed-batch or perfusion bioreactors. These reactors can
potentially be operated at high cell density without rapid cell division for a prolonged
period, with constant supply of fresh nutrient. The secreted product can be continuously
harvested from the spent medium. For cultures limited by accumulation of extra-cellular
growth inhibitors, perfusion culture is preferred. Perfusion cultures of A. officinalis
plant cells have been conducted in uniquely designed air-lift [76] and stirred-tank [34]
bioreactors for secreted protein production (Figure 3). A stirred-tank perfusion
bioreactor similar to that described in Su and Arias [34] has been used recently to
culture transgenic tobacco cells for the production of a constitutively expressed
secretory green fluorescent protein (GFP) (Su, W. and Liu, B. unpublished).
Figure 3. (A) An external-loop air-lift perfusion bioreactor (note the cell-free zone in the
upper portion of the downcomer); (B) A stirred-tank perfusion bioreactor with a cylindrical
skirt baffle; shown with the optical sensor setup for on-line monitoring of culture
fluorescence (note the cell sediment in the bottom of the bioreactor).
Perfusion bioreactors may also be operated under the fed-batch mode, with constant
recirculation of the spent medium from the cell-free zone of the reactor through a
protein recovery unit to harvest the secreted protein product (Figure 4). More
150
www.taq.ir
information on perfusion bioreactor design for plant cell cultures can be found
elsewhere [79].
Figure 4. A stirred-tank perfusion bioreactor (equipped with a skirt baffle) operated under
(A) perfusion mode with medium feeding, culture bleeding, and cell-free spent medium
removal; and (B) fed-batch mode with nutrient feeding and constant recirculation of the
spent medium through an external protein recovery unit for continuous or periodic
harvesting of the secreted protein product.
4.2. BIOREACTOR CONFIGURATIONS AND IMPELLER DESIGN
Air-lift, bubble column, and stirred-tank bioreactors have all been tested for culturing
transgenic plant cells for recombinant protein production [53], but stirred tanks are most
widely used. On the basis of time-constant/regime analysis, Doran [36] concluded that
for high-density plant cell cultures (over 30 gdw/L), mixing becomes a limiting factor in
airlift bioreactors, leading to poor oxygen transfer and heterogeneous biomass
distribution in the reactor. Another serious problem associated with pneumatically
agitated plant cell bioreactors such as the airlift and bubble columns is foaming.
Airlift/bubble columns however should work well at low to moderate biomass
concentrations. With their less complicated mechanical design, these reactors are good
candidates for low-cost bioreactors, such as the plastic-lined bubble column proposed by
Curtis and co-workers [80]. To increase reactor volumetric productivity, generally it is
preferred to operate the reactor at high cell densities, and hence stirred-tanks remain the
reactor of choice. An intricate part of designing stirred tank reactors for culturing plant
cells entails how to set the appropriate operating conditions (aeration rates, agitation
speeds, cooling/heating, etc.) so that cellular oxygen demand can be met without causing
excess foaming and shear damage to the cells. For stirred tank reactors, impeller system
is one of the most crucial elements. Doran [19] has conducted a detailed theoretical
engineering analysis of Rushton turbine (RT) and pitched blade turbines (PBT) for a
hypothetical 10 m3 stirred-tank plant cell bioreactor of standard configuration, by
concurrently considering gas dispersion, solid suspension, oxygen transfer, and shear
151
www.taq.ir
W.W. Su
damage. The analysis results were presented in flow-regime maps, which indicate that
for the RT, the minimum speed that enables complete solids and gas dispersion for
sufficient oxygen transfer is likely to cause shear damage. On the other hand, PBT
operating at the upward-pumping mode was shown in the analysis to be superior in gas
handling and solids suspension, under power input setting constrained by shear damage
considerations. Since the publication of Doran’s analysis in 1999, more studies have
been published on the hydrodynamics of upward-pumping axial-flow impellers in two
or three-phase systems, but there is no report on using such impeller in plant cell
cultures. These more recent hydrodynamics studies do support the notion that the axialflow impellers operating at an upward pumping mode is insensitive to aeration (i.e.
exhibiting low power drop upon gassing and thus not prone to impeller flooding), and is
efficient in solids suspension (i.e. minimum stirrer speed required for particle
suspension is low). However, as pointed out by Kieran [18], there are also data
indicating unfavourable mass transfer performance of upward-pumping axial-flow
impellers in viscous fermentation broths. For instance, Junker et al. [81] reported
insufficient oxygen transfer using Lightnin® A315 impeller in the up-pumping mode in
viscous Streptomyces fermentations; while the same impeller operated at the downpumping mode gave better oxygen transfer under increased broth viscosities. Nienow
and Bujalski [82] indicated that wide-blade, axial flow hydrofoils such as the A315
operated at the up-pumping mode should be considered when just physical suspension
is required or when solid-liquid reactions are rate limiting. Although not analyzed by
Doran in her work [19], due to limited hydrodynamic data available at the time, lowpower number radial flow concave blade disc impellers such as the Chemineer® CD-6
impeller have been shown to provide improved oxygen transfer (over Rushton turbines)
in Streptomyces fermentations [83]. Recently, an improved version of CD-6, called BT6, has been developed [84]. Unlike the CD-6 which has 6 symmetric concave blades,
BT-6 has six vertically asymmetric blades with the upper section of the blades longer
than the lower section (Figure 5). The BT-6 impellers exhibit very little power drop
upon gassing, even at very high flow numbers, compared with other commonly used
impeller systems, such as Rushton turbines or high solidity ratio hydrofoils. Therefore,
BT-6 is believed to be well suited for dispersing gas in reactors and fermentors where a
wide range of gas rates is required [84]. According to Chemineer® (Dayton, Ohio) [85],
the mass transfer capability of BT-6 is higher than the CD-6, on the order of 10%, and
the BT-6 is also claimed to be relatively insensitive to viscosity. These new impeller
designs (Figure 5) may indeed help improving mixing and oxygen transfer in viscous,
shear-sensitive high-density plant cell cultures, although this promise will need to be
experimentally verified first.
152
www.taq.ir
(A)
(B)
(C)
(D)
(E)
(F)
Figure 5. (A) Lightnin® A315 axial-flow impeller; (B) Lightnin® A340 up-pumping axialflow impeller; (C) Chemineer® Maxflow W axial-flow impeller; (D) Rushton disc turbine;
(E) Chemineer® CD-6 radial-flow impeller; (F) Chemineer® BT-6 radial-flow impeller.
Photograph provided courtesy of Post-Mixing.com (A, B, D, and E) [86] and Chemineer®(C
and F) [85].
4.3. ADVANCES IN PROCESS MONITORING
Research on monitoring of plant cell culture processes has largely emphasized on
detecting cell growth and related physiological parameters such as oxygen uptake rate
(OUR), carbon dioxide evolution rate (CER), and respiratory quotient (RQ). To this
end, Dalton [87] was among the first to apply off-gas analysis coupled with on-line mass
balancing to estimate the growth rate of cultured plant cells. Off-gas analysis using gas
analyzers or mass spectroscopy has also been applied by several other groups to detect
metabolic changes in plant cell cultures on-line [88-90]. Several methodologies have
been reported to directly or indirectly monitor cell concentration in plant cell suspension
culture, based on medium conductivity, osmolarity, culture turbidity (using a laser
turbidity probe), or dielectric properties (see references cited in [91]). Komaraiah et al.
[92] recently developed a multisensor array (an electronic nose) that consisted of
nineteen different metal oxide semiconductor sensors and one carbon dioxide sensor to
continuously monitor the off-gas from batch plant cell suspension cultures. Using two
pattern recognition methods, principal component analysis and artificial neural
networks, Komaraiah et al. [92] were able to analyze the multiarray responses to predict
the culture biomass concentration and formation of a secondary metabolite,
antraquinone. Availability of cell growth, OUR and CER information from on-line
measurement during bioreactor culture is useful in guiding the development of effective
substrate/inducer feeding in plant cell cultures for recombinant protein expression. Online monitoring of culture fluorescence from intrinsic fluorophores such as NAD(P)H,
153
www.taq.ir
W.W. Su
or recombinant fluorescent reporters such as GFP, can provide valuable information of
the culture metabolic states, allowing development of improved process control
strategies to increase protein production. Asali et al. [93] used NAD(P)H fluorescence
to monitor the response of starved Catharanthus roseus cells to metabolic perturbations.
Choi et al. [94] used a fibre-optic probe for on-line sensing of NAD(P)H culture
fluorescence of tobacco suspension culture and correlated the fluorescence signal to
biomass concentration. Recombinant protein product can be genetically fused with GFP
or GFP variants in a number of ways [95], allowing on-line monitoring of the
recombinant protein production by simply measuring the culture GFP fluorescence. In
addition, non-invasive detection of GFP-based sensor proteins in real time is also highly
valuable for studying the dynamics of cellular processes in plant cells that are relevant
to recombinant protein product formation. For instance, FRET (fluorescence resonance
energy transfer)-based GFP nanosensors have been developed to monitor signal
transduction and sugar transport in mammalian cells in vivo [96,97]. In the batch culture
of transgenic tobacco cells with constitutive expression of an ER-retained GFP , Liu et
al. [23] showed that culture GFP fluorescence followed closely with cell growth. A
medium feeding strategy based on culture GFP fluorescence measured off line was then
developed that resulted in improved biomass as well as GFP production in a fed-batch
culture [23]. Su et al. [95] recently demonstrated on-line monitoring of secretory GFP
production in a transgenic tobacco cell culture bioreactor using an optical light-rod
sensor. GFP culture fluorescence is a composite signal that can be influenced by factors
such as culture autofluorescence, inner filter effect (IFE), and fluorescence quenching.
These factors complicate accurate estimation of GFP concentrations from culture
fluorescence. IFE is especially problematic when using GFP in monitoring transgenic
plant cell suspension cultures, due to the aggregated nature of the cells and the high
biomass concentration in these culture systems. Reported approaches for online
compensation of IFE in monitoring culture NAD(P)H fluorescence or bioluminescence
require online measurement of biomass density or culture turbidity/optical density, in
addition to fluorescence measurement. Su et al. [98] recently developed a model-based
state observer, using the extended Kalman filter (EKF) and on-line measurement of
GFP culture fluorescence, to accurately estimate GFP concentration and other important
bioreactor states on line, while rectifying the influences of IFE and culture
autofluorescence without needing an additional biomass sensor. Software sensors,
including the use of EKF [99] and artificial neural network [100] have also been used
for monitoring biomass concentration in plant cell cultures. Zhang and Su [101]
succeeded in applying EKF coupled with simple on-line OUR measurement for
estimating the intracellular phosphate content during batch cultures of A. officinalis.
The combination of GFP-based sensing and software sensors forms a powerful tool that
can greatly advance process monitoring in transgenic plant cell cultures, allowing
development of more productive bioprocesses.
5. Future directions
In order to establish plant cell culture as a competitive host system for large-scale
commercial production of high-value recombinant proteins, the production cost has to
come down significantly. The technological/engineering advances reviewed in this
154
www.taq.ir
chapter point to many opportunities for improving recombinant protein productivity.
While further increase in productivity is expected to rely considerably on further
advances in plant molecular biology, innovative engineering solutions are equally
important to complement the molecular approaches to enhance and sustain high
productivity, as well as reducing capital and operating costs.
Acknowledgements
The author is grateful to the funding supports from the National Science Foundation
(BES97-12916 and BES01-26191), the United States Department of Agriculture
(USDA) Tropical & Subtropical Agriculture Research (TSTAR) Program (01-3413511295), and the USDA Scientific Cooperative Research Program (58-3148-9-080).
References
[1] Gomord, V. and Faye, L. (2004) Posttranslational modification of therapeutic proteins in plants. Curr.
Opin. Plant Biol. 7: 171-181.
[2] James, E. and Lee, J.M. (2001) The production of foreign proteins from genetically modified plant cells.
Adv. Biochem. Eng. Biotechnol. 72: 127-156.
[3] Crawford, K.M. and Zambryski, P.C. (1999) Plasmodesmata signaling: many roles, sophisticated statutes.
Curr. Opin. Plant Biol. 2: 382-387.
[4] Doran, P.M. (2000) Foreign protein production in plant tissue cultures. Curr. Opin. Biotechnol. 11:199204.
[5] Shin, Y.J.; Hong, S.Y.; Kwon, T.H.; Jang, Y.S. and Yang, M.S. (2003) High level of expression of
recombinant human granulocyte-macrophage colony stimulating factor in transgenic rice cell suspension
culture. Biotechnol. Bioeng. 82: 778-783.
[6] Firek, S.; Draper, J.; Owen, M.R.; Gandecha, A.; Cockburn, B. and Whitelam, G.C. (1993) Secretion of a
functional single-chain Fv protein in transgenic tobacco plants and cell suspension cultures. Plant Mol.
Biol. 23: 861-870.
[7] Fischer, R.; Liao, Y.C. and Drossard, J. (1999) Affinity-purification of a TMV-specific recombinant fullsize antibody from a transgenic tobacco suspension culture. J. Immunol. Methods 226: 1-10.
[8] Sharp, J.M. and Doran, P.M. (2001) Characterization of monoclonal antibody fragments produced by
plant cells. Biotechnol. Bioeng. 73: 338-346.
[9] Xu, H.; Montoya, F.U.; Wang, Z.; Lee, J.M.; Reeves, R.; Linthicum, D.S. and Magnuson, N.S. (2002)
Combined use of regulatory elements within the cDNA to increase the production of a soluble mouse
single-chain antibody, scFv, from tobacco cell suspension cultures. Protein Expr. Purif. 24: 384-394.
[10] Smith, M.L.; Mason, H.S. and Shuler, M.L. (2002) Hepatitis B surface antigen (HBsAg) expression in
plant cell culture: Kinetics of antigen accumulation in batch culture and its intracellular form. Biotechnol.
Bioeng. 80: 812-822.
[11] Magnuson, N.S.; Linzmaier, P.M.; Reeves, R.; An, G.; HayGlass, K. and Lee, J.M. (1998) Secretion of
biologically active human interleukin-2 and interleukin-4 from genetically modified tobacco cells in
suspension culture. Protein Expr. Purif. 13: 45-52.
[12] Kwon, T.H.; Seo, J.E.; Kim, J.; Lee, J.H.; Jang, Y.S. and Yang, M.S. (2003) Expression and secretion of
the heterodimeric protein interleukin-12 in plant cell suspension culture. Biotechnol. Bioeng. 81: 870875.
[13] James, E.A.; Wang, C.; Wang, Z.; Reeves, R.; Shin, J.H.; Magnuson, N.S. and Lee, J.M. (2000)
Production and characterization of biologically active human GM-CSF secreted by genetically modified
plant cells. Protein Expr. Purif. 19: 131-138.
[14] Francisco, J.A.; Gawlak, S.L.; Miller, M.; Bathe, J.; Russell, D.; Chace, D.; Mixan, B.; Zhao, L.; Fell,
H.P. and Siegall, C.B. (1997) Expression and characterization of bryodin 1 and a bryodin 1-based singlechain immunotoxin from tobacco cell culture. Bioconjug. Chem. 8: 708-713.
155
www.taq.ir
W.W. Su
[15] Terashima, M.; Murai, Y.; Kawamura, M.; Nakanishi, S.; Stoltz, T.; Chen, L.; Drohan, W.; Rodriguez,
R.L. and Katoh, S. (1999) Production of functional human alpha 1-antitrypsin by plant cell culture. Appl.
Microbiol. Biotechnol. 52: 516-523.
[16] Trexler, M.M.; McDonald, K.A. and Jackman, A.P. (2002) Bioreactor production of human alpha(1)antitrypsin using metabolically regulated plant cell cultures. Biotechnol. Prog. 18: 501-508.
[17] Fischer, R.; Emans, N.; Schuster, F.; Hellwig, S. and Drossard, J. (1999) Towards molecular farming in
the future: using plant-cell-suspension cultures as bioreactors. Biotechnol. Appl. Biochem. 30 (Pt 2):
109-112.
[18] Kieran, P.M. (2001) Bioreactor design for plant cell suspension cultures. In: Cabral, J.; Mota, M. and
Tramper, J. (Eds.) Multiphase bioreactor design. Taylor and Francis, London; pp. 391-426.
[19] Doran, P.M. (1999) Design of mixing systems for plant cell suspensions in stirred reactors. Biotechnol.
Prog. 15: 319-335.
[20] Wetzstein, H. and He, Y. (2000) Anatomy of plant cells. In: Spier, R (Ed.), Encyclopedia of cell
technology. Wiley, New York; pp. 24-31.
[21] Kwon, T.; Kim, Y.; Lee, J. and Yang, M. (2003) Production and secretion of biologically active human
granulocyte-macrophage colony stimulating factor in transgenic tomato suspension cultures. Biotechnol.
Letters 25: 1571-1574.
[22] Kwon, S.; Jo, S.; Lee, O.; Choi, S.; Kwak, S. and Lee, H. (2003) Transgenic ginseng cell lines that
produce high levels of a human lactoferrin. Planta Medica 69: 1005-1008.
[23] Liu, S.; Bugos, R.C.; Dharmasiri, N. and Su, W.W. (2001) Green fluorescent protein as a secretory
reporter and a tool for process optimization in transgenic plant cell cultures. J. Biotechnol. 87: 1-16.
[24] Su, W.; Lei, F. and Su, L. (1993) Perfusion strategy for rosmarinic acid production by Anchusa
officinalis. Biotechnol. Bioeng. 42: 884-890.
[25] Hibino, K. and Ushiyama, K. (1999) Commercial production of ginseng by plant tissue culture
technology. In: Far, T.; Singh, G. and Curtis, W. (Eds.) Plant Cell and Tissue Culture for the Production
of Food Ingredients. Kluwer Academic Publisher, New York; pp. 215-224.
[26] Chattopadhyay, S.; Farkya, S.; Srivastava, A. and Bisaria, V. (2002) Bioprocess considerations for
production of secondary metabolites by plant cell suspension cultures. Biotechnol. Bioproc. Eng. 7: 138149.
[27] KeEler, M.; ten Hoopen, H. and Furusaki, S. (1999) The effect of aggregate size on the production of
ajmalicine and tryptamine in Catharanthus roseus suspension culture. Enz. Microbial. Technol. 24: 308315.
[28] Su, W. (1995) Bioprocessing technology for plant cell suspension cultures. Appl. Biochem. Biotechnol.
50: 189-230.
[29] Danna, K. (2001) Production of cellulases in plants for biomass conversion. Recent Adv. in
Phytochemistry 35: 205-231.
[30] Cosgrove, D.J. (1997) Relaxation in a high-stress environment: the molecular bases of extensible cell
walls and cell enlargement. Plant Cell 9: 1031-1041.
[31] Vissenberg, K.; Feijo, J.A.; Weisenseel, M.H. and Verbelen, J.P. (2001) Ion fluxes, auxin and the
induction of elongation growth in Nicotiana tabacum cells. J. Exp. Bot. 52: 2161-2167.
[32] Joubes, J.; De Schutter, K.; Verkest, A.; Inze, D. and De Veylder, L. (2004) Conditional, recombinasemediated expression of genes in plant cell cultures. Plant J. 37: 889-896.
[33] Curtis, W. and Emery, A. (1993) Plant cell suspension culture rheology. Biotechnol. Bioeng. 42: 520526.
[34] Su, W. and Arias, R. (2003) Continuous perfusion plant cell culture: Bioreactor characterization and
secreted enzyme production. J. Biosci. Bioeng. 95: 13-20.
[35] Kieran, P.M.; MacLoughlin, P.F. and Malone, D.M. (1997) Plant cell suspension cultures: some
[36] Doran, P. (1993) Design of reactors for plant cells and organs. Adv. Biochem. Eng. Biotechnol. 48:115168.
[37] Abdullah, M.; Ariff, A.; Marziah, M.; Ali, A. and Lajis, N. (2000) Strategies to overcome foaming and
wall growth during the cultivation of Morinda elliptica cell suspension culture in a stirred-tank
bioreactor. Plant Cell Tissue Org. Cult. 60: 205-212.
[38] Wongsamuth, R. and Doran, P. (1997) The filtration properties of Atropa belladonna plant cell
suspensions; effect of hydrodynamic shear and elevated carbon dioxide levels on culture and filtration
properties. J. Chem. Technol. Biotechnol. 69: 15-26.
156
www.taq.ir
[39] Howell, J.; Chi, C. and Pawlowsky, U. (1972) Effect of wall growth on scale-up problems and dynamic
operating characteristics of the biological reactor. Biotechnol. Bioeng. 14: 253-265.
[40] Kawase, Y. and Moo-Young, M. (1990) The effect of antifoam agents on mass transfer in bioreactors.
Bioprocess Eng. 5: 169-173.
[41] Meijer, J.; ten Hoopen, H.; Luyben, K. and Libbenga, K. (1993) Effects of hydrodynamic stress on
cultured plant cells: A literature survey. Enz. Microbial Technol. 15: 234-238.
[42] Kieran, P.M.; Malone, D.M. and MacLoughlin, P.F. (2000) Effects of hydrodynamic and interfacial
forces on plant cell suspension systems. Adv. Biochem. Eng. Biotechnol. 67: 139-177.
[43] Dunlop, E.; Namdev, P. and Rosenberg, M. (1994) Effect of fluid shear forces on plant cell suspensions.
Chemical Eng. Sci. 49: 2263-2276.
[44] Sowana, D.; Williams, D.; Dunlop, E.; Dally, B.; O'Neill, B. and Fletcher, D. (2001) Turbulent shear
stress effects on plant cell suspension cultures. Trans. Chem.E 79: 867-875.
[45] Sowana, D.; Williams, D.; O'Neill, B. and Dunlop, E. (2002) Studies of the shear protective effects of
Pluronic F-68 on wild carrot cell cultures. Biochemical Eng. J. 12: 165-173.
[46] MacLoughlin, P.F.; Malone, D.M.; Murtagh, J.T. and Kieran, P.M. (1998) The effects of turbulent jet
flows on plant cell suspension cultures. Biotechnol. Bioeng. 58: 595-604.
[47] Namdev, P.K. and Dunlop, E.H. (1995) Shear sensitivity of plant cells in suspension. Appl. Biochem.
[48] Han, R. and Yuan, Y. (2004) Oxidative burst in suspension culture of Taxus cuspidata induced by a
laminar shear stress in short-term. Biotechnol. Progress 20: 507-513.
[49] Yahraus, T.; Chandra, S.; Legendre, L. and Low, P.S. (1995) Evidence for a mechanically induced
oxidative burst. Plant Physiol. 109: 1259-1266.
[50] Shinmyo, A.; Shoji, T.; Bando, E.; Nagaya, S.; Nakai, Y.; Kato, K.; Sekine, M. and Yoshida, K. (1998)
Metabolic engineering of cultured tobacco cells. Biotechnol. Bioeng. 58: 329-332.
[51] Koroleva, O.A.; Tomlinson, M.; Parinyapong, P.; Sakvarelidze, L.; Leader, D.; Shaw, P. and Doonan,
J.H. (2004) CycD1, a Putative G1 Cyclin from Antirrhinum majus, accelerates the cell cycle in cultured
tobacco BY-2 Cells by enhancing both G1/S entry and progression through S and G2 phases. Plant Cell
16: 2364-2379.
[52] Cockcroft, C.E.; den Boer, B.G.; Healy, J.M. and Murray, J.A. (2000) Cyclin D control of growth rate in
plants. Nature 405: 575-579.
[53] Gao, J. and Lee, J. (1992) Effect of oxygen supply on the suspension culture of genetically modified
tobacco cells. Biotechnol. Progress 8: 285-290.
[54] Cooney, C.; Wang, D. and Mateles, R. (1969) Measurement of heat evolution and correlation with
oxygen consumption during microbial growth. Biotechnol. Bioeng. 11: 269-281.
[55] Hashimoto, T. and Azechi, S. (1988) Bioreactors for large-scale culture of plant cells. In: Bajaj, Y.P.S.
(Ed.), Biotechnology in Agriculture and Forestry. Springer, Berlin; pp. 104-122.
[56] Farres, J. and Kallio, P.T. (2002) Improved cell growth in tobacco suspension cultures expressing
Vitreoscilla hemoglobin. Biotechnol. Progress 18: 229-233.
[57] Igamberdiev, A.U.; Seregelyes, C.; Manac'h, N. and Hill, R.D. (2004) NADH-dependent metabolism of
nitric oxide in alfalfa root cultures expressing barley hemoglobin. Planta 219: 95-102.
[58] Shiao, T.L.; Ellis, M.H.; Dolferus, R.; Dennis, E.S. and Doran, P.M. (2002) Overexpression of alcohol
dehydrogenase or pyruvate decarboxylase improves growth of hairy roots at reduced oxygen
concentrations. Biotechnol. Bioeng. 77: 455-461.
[59] Suehara, K.; Takao, A.; Nakamura, K.; Uozumi, N. and Kobayashi, T. (1996) Optimal expression of
GUS gene from methyl jasmonate-inducible promoter in high density culture of transformed tobacco cell
line BY-2. J. Ferment. Bioeng. 82: 51-55.
[60] Yoshida, K.; Kasai, T.; Garcia, M.R.; Sawada, S.; Shoji, T.; Shimizu, S.; Yamazaki, K.; Komeda, Y. and
Shinmyo, A. (1995) Heat-inducible expression system for a foreign gene in cultured tobacco cells using
the HSP18.2 promoter of Arabidopsis thaliana. Appl. Microbiol. Biotechnol. 44: 466-472.
[61] Uozumi, N.; Inoue, Y.; Yamazaki, K. and Kobayashi, T. (1994) Light activation of expression associated
with the tomato rbcS promoter in transformed tobacco cell line BY-2. J. Biotechnol. 36: 55-62.
[62] Nara, Y.; Kurata, H.; Seki, M. and Taira, K. (2000) Glucocorticoid-induced expression of a foreign gene
by the GVG system in transformed tobacco BY-2 cells. Biochemical Eng. J. 6: 185-191.
[63] Kim, K.Y.; Kwon, S.Y.; Lee, H.S.; Hur, Y.; Bang, J.W. and Kwak, S.S. (2003) A novel oxidative stressinducible peroxidase promoter from sweet potato: molecular cloning and characterization in transgenic
tobacco plants and cultured cells. Plant Mol. Biol. 51: 831-838.
157
www.taq.ir
W.W. Su
[64] Boetti, H.; Chevalier, L.; Denmat, L.A.; Thomas, D. and Thomasset, B. (1999) Efficiency of physical
(light) or chemical (ABA, tetracycline, CuSO4 or 2-CBSU)-stimulus-dependent gus gene expression in
tobacco cell suspensions. Biotechnol. Bioeng. 64: 1-13.
[65] Nagaya, S.; Nakai, Y., Kato, K; Sekine, M.; Yoshida, K. and Shinmyo, A. (2000) Isolation of growthphase-specific promoters from cultured tobacco cells. J. Biosci. Bioeng. 89: 231-235.
[66] Fischer, R.; Stoger, E.; Schillberg, S.; Christou, P. and Twyman, R.M. (2004) Plant-based production of
biopharmaceuticals. Curr. Opin. Plant Biol. 7: 152-158.
[67] Bateman, K.; Congiu, M.; Tregear, G.; Clarke, A. and Anderson, M. (1997) Bacitracin significantly
reduces degradation of peptides in plant cell cultures. Biotechnol. Bioeng. 53: 226-231.
[68] James, E.; Mills, D. and Lee, J. (2002) Increased production and recovery of secreted foreign proteins
from plant cell cultures using an affinity chromatography bioreactor. Biochemical Eng. J. 12: 205-213.
[69] Goddijn, O. and Pen, J. (1995) Plants as bioreactors. TIBTECH 13: 379-387.
[70] Gallie, D. and Walbot, V. (1992) Identification of the motifs within the tobacco mosaic virus 5'-leader
responsible for enhancing translation. Nucleic Acids Res. 20: 4631-4638.
[71] Day, C.; Lee, E.; Kobayashi, J.; Holappa, L.; Albert, H. and Ow, D. (2000) Transgene integration into
the same chromosome location can produce alleles that express at a predictable level, or alleles that are
differentially silenced. Genes Dev. 14: 2869-2880.
[72] Hilleren, P. and Parker, R. (1999) Mechanisms of mRNA surveillance in eukaryotes. Ann. Rev. Genet.
33: 229-260.
[73] Spiker, S. and William, F. (1996) Nuclear matrix attachment regions and transgene expression in plants.
Plant Physiol. 110: 15-21.
[74] Voinnet, O.; Pinto, Y. and Baulcombe, D. (1999) Suppression of gene silencing: A general strategy used
by diverse DNA and RNA viruses of plants. Proc. Natl. Acad. Sci. USA 96: 14147-14152.
[75] Verdelhan des Molles, D.; Gomord, V.; Bastin, M.; Faye, L. and Courtois, D. (1999) Expression of a
carrot invertase gene in tobacco suspension cells cultivated in batch and continuous culture conditions. J.
Biosci. Bioeng. 87: 302-306.
[76] Su, W.; He, B.; Liang, H. and Sun, S. (1996) A perfusion air-lift bioreactor for high density plant cell
cultivation and secreted protein production. J. Biotechnol. 50: 225-233.
[77] Drapeau, D.; Blanch, H.W. and Wilke, C.R. (1987) Economic assessment of plant cell culture for the
production of ajmalicine. Biotechnol. Bioeng. 30: 946-953.
[78] Terashima, M.; Ejiri, Y.; Hashikawa, N. and Yoshida, H. (2001) Utilization of an alternative carbon
source for efficient production of human alpha(1)-antitrypsin by genetically engineered rice cell culture.
Biotechnol. Prog. 17: 403-406.
[79] Su, W. (2000) Perfusion bioreactors. In: Spier, R. (Ed.), Encyclopedia of Cell Technology. Wiley, New
York; pp. 230-242.
[80] Hsiao, T.Y.; Bacani, F.T.; Carvalho, E.B. and Curtis, W.R. (1999) Development of a low capital
investment reactor system: application for plant cell suspension culture. Biotechnol. Prog. 15: 114-122.
[81] Junker, B.; Stanik, M.; Barna, C.; Salmon, P. and Buckland, B. (1998) Influence of impeller type on
mass transfer in fermentation vessels. Bioproc.Eng. 19: 403-413.
[82] Nienow, A.W. and Bujalski, W. (2002) Recent studies on agitated three-phase (gas-solid-liquid) systems
in the turbulent regime. Chemical Eng. Res. Design 80: 832-838.
[83] Junker, B.H.; Mann, Z. and Hunt, G. (2000) Retrofit of CD-6 (Smith) impeller in fermentation vessels.
Appl. Biochem. Biotechnol. 89: 67-83.
[84] Pinelli, D.; Bakker, A.; Myers, K.J.; Reeder, M.F.; Fasano, J. and Magelli, F. (2003) Some features of a
novel gas dispersion impeller in a dual-impeller configuration. Chemical Eng. Res. Design 81: 448-454.
[85] Anon. (2002) http://www.chemineer.com/impellers.php.
[86] Csiszar, P. (2004) http://www.postmixing.com/.
[87] Dalton, C. (1985) Application of gas analysis to continuous culture. In: Neumann, K.; Barz, W. and
Reinhard, E. (Eds.) Primary and secondary metabolism of plant cell cultures. Springer, Berlin; pp. 58-65.
[88] Bond, P.; Fowler, M. and Scragg, A. (1988) Growth of Catharanthus roseus cell suspensions in
bioreactors: on-line analysis of oxygen and carbon dioxide levels in inlet and outlet gas streams.
Biotechnol. Lett 10: 713-718.
[89] Nikolova, P.; Moo-Young, M. and Legge, R. (1991) Application of process mass spectroscopy to the
detection of metabolic changes in plant tissue culture. Plant Cell Tissue Org. Cult. 25: 219-224.
[90] Zhong, J.; Konstantinov, K. and Yoshida, T. (1994) Computer-aided on-line monitoring of physiological
variables in suspended cell cultures of Perilla frutescens in a bioreactor. J. Ferment. Bioeng. 77: 445-447.
158
www.taq.ir
[91] Zhong, J. (2001) Biochemical engineering of the production of plant-specific secondary metabolites by
cell suspension cultures. Adv. Biochemical Eng. Biotechnol. 72: 1-26.
[92] Komaraiah, P.; Navratil, M.; Carlsson, M.; Jeffers, P.; Brodelius, M.; Brodelius, P.E.; Kieran, P.M. and
Mandenius, C.F. (2004) Growth behaviour in plant cell cultures based on emissions detected by a
multisensor array. Biotechnol. Prog. 20: 1245-1250.
[93] Asali, E.C.; Mutlmmma, R. and Humphrey, A.E. (1992) Use of NAD(P)H-fluorescence for monitoring
the response of starved cells of Catharanthus roseus in suspension to metabolic perturbations. J.
[94] Choi, J.; Park, Y.; shin, C.; Kim, D. and Lee, W. (1995) Analysis of culture fluorescence by a fiber-optic
sensor in Nicotiana tabacum plant cell culture. Korean J. Chemical Eng. 12: 528-534.
[95] Su, W.W.; Guan, P. and Bugos, R.C. (2004) High-level secretion of functional green fluorescent protein
from transgenic tobacco cell cultures: characterization and sensing. Biotechnol. Bioeng. 85: 610-619.
[96] Miyawaki, A.; Liopis, J.; Heim, R.; McCaffery, J.M.; Adams, J.A.; Ikura, M. and Tsien, R.Y. (1997)
Fluorescent indicators for Ca2+ based on green fluorescent proteins and calmodulin. Nature 388: 882-887.
[97] Fehr, M.; Frommer, W.B. and Lalonde, S. (2002) Visualization of maltose uptake in living yeast cells by
fluorescent nanosensors. Proc. Natl. Acad. Sci. USA 99: 9846-9851.
[98] Su, W.; Liu, B.; Lu, W.; Xu, N.; Du, G. and Tan, J. (2004) Observer-based online compensation of inner
filter effect in monitoring fluorescence of GFP-expressing plant cell cultures. (under publication).
[99] Albiol, J.; Robuste, J.; Casas, C. and Poch, M. (1993) Biomass estimation in plant cell cultures using
extended Kalman filter. Biotechnol. Progress 9: 174-178.
[100] Albiol, J.; Campmajo, C.; Casas, C. and Poch, M. (1995) Biomass estimation in plant cell cultures: a
neural network approach. Biotechnol. Progress 11: 88-92.
[101] Zhang, J. and Su, W. (2002) Estimation of intracellular phosphate content in plant cell cultures using an
extended Kalman filter. J. Biosci. Bioeng. 94: 8-14.
159
www.taq.ir
TYPES AND DESIGNS OF BIOREACTORS FOR HAIRY ROOT CULTURE
YONG-EUI CHOI1, YOON-SOO KIM2 AND KEE-YOEUP PAEK3
1
Department of Forestry, College of Forest Sciences, Kangwon National
University, Chunchon 200-701, Kangwon-do, Korea – Fax: 82-33-2528310 – Email: [email protected]
2
Korea Ginseng Institute, Chung-Ang University, Ansung-shi, Kyunggido, Korea – Fax: 82-31-676-6544 – Email: [email protected]
3
Research Centre for the Development of Advanced Horticultural
Technology, Chungbuk National University, Cheongju 361-763, KoreaFax: 82-43-272-5369 – Email: [email protected]
1. Introduction
Plants synthesize a wide range of secondary metabolites such as alkaloids,
anthocyanins, flavonoids, quinins, lignans, steroids, and terpenoids, which play a major
role in the adaptation of plants to their environment. The secondary metabolites have
been used as food additives, drugs, dyes, flavours, fragrances, and insecticides. Such
chemicals are extracted and purified from naturally grown plants. However, production
of secondary metabolites from plants is not always satisfactory. It is often restricted to a
limited species or genus, and geographically to a specific region. Many important
medicinal plants were endangered by overexploitation. Some plants are difficult to
cultivate and grow very slowly or are endangered in their natural habitats. The
biotechnological approach by utilizing plant cell and organ culture system can offer an
opportunity to produce the secondary metabolites. Plant materials via in vitro culture
are produced with high uniformity regardless of geographical and seasonal limitations
and environmental factors. However, there are many problems in the production of
metabolites by plant cell and organ culture technology due to the high cost to natural
counterparts, and the low yield of metabolites in cultured plant cells. Although there are
many efforts for establishing the cell and organ culture systems, application in the
commercial production of pharmaceuticals is limited to a few examples only.
Production of shikonin from the cell culture of Lithospermum erythrorhizon [1,2], taxol
from Taxus baccata [3] and berberine from Coptis japonica [4] was reached for the
application for industrialization. The main problem using cell suspension culture is a
low product yield and instability of the cell lines [5].
The secondary metabolites can be produced by developed organ and plantlets [6,7].
An alternative method for the production of plant materials for secondary metabolite
production is the culture of shoots, roots, or whole plants. However, the organ culture
tends to grow slowly and renders the difficulty of the large-scale cultivation compared
161
© 2008 Springer.
www.taq.ir
Y-E. Choi, Y-S. Kim and K-Y. Paek
to cell culture. Agrobacterium rhizogenes-transformed hairy roots synthesize the same
component as does the roots of the intact plants and have a fast growth property in
hormone-free medium. Many efforts have been made to commercialize the plant
metabolites via a bioreactor culture of hairy roots. The bioreactor for microorganism
fermentation (stirred tank bioreactor) is unsuitable for the mass production of hairy
roots because of strong shear stress. Therefore, various types of bioreactor systems were
designed and evolved to enhance the productivity and the bioprocess. Among them,
airlift, bubble column, and liquid-dispersed bioreactor are largely adopted for the hairy
root culture because of the low shear stress and the simplicity of their design and
construction. Significant progress has been made in biotechnology and bioprocess for
the large-scale culture of hairy roots. In this chapter, we focus on the recent technology
covering the bioreactor culture systems, such as the shape of bioreactor, aeration
condition, and introduce the large-scale production of ginseng hairy-like roots for
commercialization.
2. Advantage of hairy root cultures
Normally, adventitious root cultures need an exogenous phytohormone supply and grow
very slowly. Hairy roots can be produced by transformation with the soil bacterium
Agrobacterium rhizogenes, resulting in the so-called hairy roots disease [8]. Long-term
genetic and biosynthetic stability was noted from this type of culture [9,10]. In addition,
they produce similar secondary metabolites to the normal roots and much higher levels
than do cell cultures [6,11,12]. Therefore, hairy roots can offer a valuable source of
root-derived secondary metabolites that are useful as pharmaceuticals, cosmetics, and
food additives. Transformed roots of many plant species have been widely studied for
the in vitro production of secondary metabolites [13,14].
Another interesting strategy of hairy root cultures is the genetic engineering of
secondary metabolism by introducing useful genes. Enhanced production of alkaloid
nicotine by the introduction of ornithine decarboxylase into Nicotiana rustica was
reported [15]. The hairy roots of Atropa belladona overexpressing hyoscyamine 6-betahydroxylase (H6H) gene isolated from Hyoscyamus niger produced high amounts of
scopolamine [16]. In Hyoscyamus niger hairy root cultures, overexpression of genes
encoding both putrescine N-methyltransferase (PMT) and the downstream enzyme
hyoscyamine-6-beta-hydroxylase (H6H) resulted in the enhanced scopolamine
biosynthesis [17]. Hairy root cultures of Datura metel overexpressing the SAM Nmethyltransferase (PMT) gene encodes for putrescine, which accumulated higher
amounts of tropane alkaloids (hyoscyamine and scopolamine) than do the control hairy
roots [18]. The transgenic hairy roots by introducing the genes regulating secondary
metabolism will provide an effective approach for efficient and large-scale commercial
production of secondary metabolite production.
3. Induction of hairy roots
Hairy roots are induced from the transfer and integration of the genes of Ri plasmid of
Agrobacterium rhizogenes [8]. Integration of a DNA segment (T-DNA) of Ri-plasmid
162
www.taq.ir
www.taq.ir
Types and designs of bioreactors for hairy root culture
into the host plant genome results in the active proliferation of hairy roots [8]. The Ri
plasmids are grouped into two main classes: agropine and mannopine type strains [19].
The agropine type strains contain both the TL (about 15-20 kb) and TR (about 8-20 kb)
region in their Ri plasmid are more virulent than mannopine strains, and are therefore
more often used for the establishment of hairy root cultures [20]. Agrobacterium
rhizogenes A4 type (A4, ATCC, 15834, 1855, TR105, etc) can synthesize both agropine
and mannopine. Agrobacterium rhizogenes 8196 type (TR7, TR101, etc.) synthesize the
mannopine only.
The vir region comprises about 35 kb in the Ri plasmid, and encodes six
transcriptional loci: vir A, B, C, D, E, and G, which have important functions in gene
transfer. Transcription of the vir region is induced by various phenolic compounds such
as acetosyringone [21]. Acetosyringone or related compounds have been reported to
increase the frequency of Agrobacterium mediated transformations in a number of plant
species [22], especially for recalcitrant monocotyledonous plant species [23]. Various
sugars also act synergistically with acetosyringone to induce a high level of vir gene
expression [24,25].
In the agropine Ri plasmid T-DNA is referred to as left T-DNA (TL-DNA) and right
T-DNA (TR-DNA) [26]. Genes involved in agropine and auxin syntheses are located in
the TR DNA region. Genes of Ri TL-DNA such as rolA, rolB, rolC and rolD stimulate
hairy root differentiations under the influence of endogenous auxin synthesis [27]. TDNA analysis in hairy roots reveals that TL and TR-DNAs exist in random manners
either as distinct inserts, or as a single and continuous insert including the region
between TL and TR on pRi 15834 [28]. Sequencing of genomic DNA/T-DNA junctions
in hairy roots reveals that genomic DNA at the cleavage sites are usually intact, whereas
donor T-DNA ends are often resected, as are found in random T-DNA inserts. Batra et
al. [29] reported that growth and terpenoid indole alkaloid production in Catharanthus
roseus hairy root clones is related to left and right-termini-linked Ri T-DNA gene
integration. Therefore, each hairy root line shows different morphology and growth
pattern together with different biosynthetic capability of secondary metabolites.
4. Large-scale culture of hairy roots
Generally, the hairy root culture in bioreactors is focused on both secondary metabolites
production via the biomass growth of root tissues. Growth of hairy roots and production
of secondary metabolites is controlled by the genetic characteristics of plant species, and
they are strongly influenced by physical and chemical culture conditions such as the
types of culture vessels, composition and concentration of macro and micro-element,
concentration of carbon sources, pH, light, and temperature etc. In hairy root culture
systems, biomass growth is achieved due to a series of two characteristic growths: the
lateral root primordium formation on parent root segments and their elongation [30]. In
comparison to a cell suspension culture, the growth of hairy roots in liquid medium
results in the packed root mass playing an inhibitory role in fluid flow and limiting
oxygen availability [31]. In addition, the roots hairs play a detrimental role for the
growth in a liquid environment because they induce the stagnation of fluid flow and
limit the availability of oxygen [31]. Therefore, the morphological character of hairy
roots and oxygen supply are primary factors for designing and optimizing the culture
163
www.taq.ir
condition of hairy roots [32,33]. To achieve successfully a scale-up, reactor types and
assessments of reactor performance must be considered to minimize the problems,
which will be encountered during the scale-up. In the case of the Erlenmeyer flask
culture, it is very difficult to modify the culture environment within flasks and is used
for only small-scale culture due to the limited air supply. A bioreactor fitted with
controllers for air supply, pH, temperature etc. is mainly utilized for the large-scale
culture of hairy roots. Various configurations of hairy root bioreactors such as the
stirred tank, airlift, bubble column, liquid-dispersed bioreactor have been designed for
hairy root cultures [14,34]. Therefore, we introduce the cultures of well-known
bioreactors for the production of hairy roots and recent advances on the bioreactor
culture technology for large-scale production of hairy roots.
4.1. STIRRED TANK REACTOR
In this type of bioreactor, mortar-derived impeller or turbine blades regulate aeration
and medium currency. This reactor is widely adopted for microorganism, fermentation
and plant cell culture. Temperature, pH, amount of dissolved oxygen, and nutrient
concentration can be better controlled within this reactor than in other type of reactors.
In general, the impellers used in this reactor produce a high-shear stress compared to
other types [35-37]. For hairy roots culture, the impeller must be operated with
restricted power input and speed to minimize the shear stress. Ways of improving
impeller performance by modifying internal reactor geometry have been designed [3840]. In the hairy root culture of Catharanthus trichophyllus, hairy root line cultures in
stirred bioreactor showed a similar alkaloid composition to normal root [41]. The
cultivation of Swertia chirata hairy roots in a 2-L stirred-tank bioreactor was successful
only with a stainless-steel mesh fitted inside the culture vessel for immobilization of the
roots [42]. In the Panax ginseng hairy root culture, the growth of roots in a stirred
bioreactor in which stainless-steel mesh fitted in culture vessel was about three times as
high as in the flask cultivation [43].
4.2. AIRLIFT BIOREACTORS
In the airlift bioreactor, both liquid currency and aeration are driven by externally
supplied air. This reactor is advantageous for the culture of plant cells and organs those
are sensitive to shear stress. However, this reactor is not suitable for high-density culture
because of insufficient mixing process inside the reactor. In 2.5-L hairy root culture of
Pueraria phaseoloides, puerarin accumulation is 200 times as much as in a 250 ml
shake flask culture [44]. In the hairy root culture of Astragalus membranaceus, both the
dry weight of hairy roots and astragaloside IV from a 30-L airlift bioreactor were higher
than the yields from a 10-L bioreactor [45]. In the Panax ginseng hairy root culture, the
growth of roots in both the bubble column and the stirred bioreactor was about three
times as high as in the flask cultivation [46]. Hairy roots growth was about 55-fold of
inoculums after 39 d in a 5-L airlift bioreactor and about 38-fold of inoculums after 40 d
in a 19-L airlift bioreactor [43].
164
www.taq.ir
4.3. BUBBLE COLUMN REACTOR
The bubble column reactor is one of simplest types of reactors and is easy to scale-up.
Its disadvantage is the undefined flow pattern inside the reactor resulting into nonuniform mixing. Like an airlift bioreactor, the bubbles in a bubble column create less
shear stress compared to other stirred types, so that it is useful for organized structures
such as hairy roots. In this case, the bubbling rate needs to be gradually increased with
the growth of hairy roots. However, at a high tissue density level, the bubble column
has been observed to reduce growth performance [47]. In hairy root culture of Solanum
tuberosum in a 15-L bubble column, stagnation and channelling of gas through the bed
of growing roots exists, however, the gas-liquid interface is not the dominant resistance
factor to oxygen mass transfer, and the oxygen uptake of growing tips increase with the
oxygen tension of the medium [48]. The growth and production of hyoscyamine and
scopolamine in the culture of hairy roots of Datura metel was enhanced by the
treatment of permeabilizing agent Tween 20 in an airlift bioreactor with root anchorage
[49]. In hairy root cultures of Hyoscyamus muticus accumulated tissue mass in
submerged air-sparged reactors was 31% of gyratory shake-flask controls [50]. They
reported that impaired oxygen transfer due to channelling and stagnation of the liquid
phase are the apparent causes of poor growth [50]. Inclusion of polyurethane foam in
the vessel of air-sparged bioreactor reduces the entrapping of gas by hairy roots, which
improve biomass and alkaloid production [51]. In Artemisia annua hairy root culture,
the bubble column reactor was superior to mist reactors for the biomass concentration
[52,53]. Souret et al. [53] examined the difference between the two types of bioreactors,
a mist reactor and a bubble column reactor. Mist reactors produce significantly more
artemisinin, while bubble column reactors produce greater biomass. The roots grown in
shake flasks contain a negligible amount of artemisinin. The high-density culture of red
beet hairy roots was obtained by a radial flow reactor, which consists of a cylindrical
vessel with a radial flow of medium [54].
4.4. LIQUID-DISPERSED BIOREACTOR
The reactors used for hairy root culture can be classified as either liquid-phase or gasphase. Liquid-dispersed reactor is advantageous both for sufficient oxygen supply to
roots and for a low shear stress environment compared with reactors in which the roots
remained submerged in a liquid medium [50]. In liquid-dispersed reactors, roots are
exposed to ambient air, or gas mixture, and the nutrient liquid, which is dispersed as
spray or mist onto the top of the root bed [52,55]. The sprayed liquid and mist are
drained from the bottom of the bioreactor to a reservoir and is re-circulated. The degree
of distribution of liquid varies according to the mechanism of liquid delivery at the top
of the reactor chamber. Various types of liquid-dispersed reactors are developed for the
hairy root culture. Mist or nutrient mist [56-59], droplet [52,59], trickle-bed or tricking
film [57,60], and drip-tube [61] are reported. In these bioreactors, certain types of
configurations to internal support of roots such as glass beads, rasching rings, steel wire
scaffolding, polyurethane foam, horizontal mesh trays, and cylindrical stainless steel
mesh are invented [52,57,59-61]. Cichorium intybus hairy roots grown in an acoustic
mist bioreactor produce nearly twice as much aesculin as compared to roots grown in
bubble column and nutrient sprinkle bioreactors [62]. Artemisia annua hairy roots
165
www.taq.ir
grown in nutrient mist reactors produce nearly three times as much artemisinin as roots
grown in bubble column reactors [63], and the authors suggest that higher levels of
artemisinin in roots grown in the mist reactors are due to a response to the increased
osmotic strength of the medium within the mist reactor, the medium becomes
concentrated due to water evaporation [63]. In contrast to artemisinin accumulation in
Artemisia annua hairy roots, the mist reactor accumulates lower biomass than does the
bubble column reactor due to insufficient nutrient availability [52].
5. Commercial production of Panax ginseng roots via balloon type bioreactor
Panax ginseng has been used for important Oriental medicine since ancient time, owing
to its tonic properties. The ginseng root contains terperpenoid saponins, referred to as
ginsenosides. Cultivation of ginseng requires at least more than four years under shade
condition and also requires the careful control of disease. Cell and organ culture
technology have been developed for the alternative production of ginseng raw materials
and secondary metabolites. The ginseng cell culture has been applied to the production
of useful secondary metabolites [64,65]. Hormone-independent embryogenic cells are
induced and cultivated via a bioreactor [66,67]. The cell suspensions produced from
pilot scale culture have been commercialized into various ginseng tea and tonic
beverages by Nitto Denko Co., Japan. [68].
Hairy roots provide an efficient way of biomass production due to fast growth and
displays high biosynthetic capabilities that are comparable to those of natural roots [6,
11,12]. There are many publications on the hairy root culture of ginseng [43,69].
However, hairy roots are still not well utilized for the production of health food and
need further analysis for the safety of proteins and compounds expressed by introduced
genes of T-DNA. Recently, hairy-like adventitious roots culture without transformation
with Agrobacterium rhizogenes was reported [70,71]. Induction and growth of hairylike adventitious roots is achieved from initial root explants by exogenous auxin supply,
which is direct motive for the mass production of ginseng roots for commercial scale.
Son et al. [71] designed a balloon-type bubble bioreactor (BTBB) (Figures 1, 2A),
which is superior for biomass growth than the bubble column bioreactor, and stirred
tank bioreactor in cell culture of Taxus cuspidata [72], Beta vulgaris hairy roots [73],
ginseng hairy root [74] and adventitious root culture [75]. The fresh weight of ginseng
hairy-like adventitious root culture in 20-L BTBB was three-times higher than that of
the stirred tank bioreactor [71]. The maximum biomass production of 2.2 kg fresh
weight in 20-L bioreactor was obtained after 42 days after inoculation of 240 g [76]. In
mountain ginseng cell line maintained by CBN Biotech Co., Korea, biomass growth of
ginseng roots is reached to 30-fold of inoculums after 42 days of culture (Table 1).
166
www.taq.ir
Figure 1. Actively growing ginseng hairy roots in 20-L balloon-type bubble bioreactor after
42 days of culture. Photograph provided by Son SH of VitroSys Co., Korea.
Table 1. Growth and saponin accumulation of adventitious ginseng roots after 42 days of
culture in 5, 20, 500 and 1,000-L balloon-type bubble bioreactors.
Working
volume (L)
Inoculums Fresh Wt. Dry Wt.
(g)
(g)
(g)
Saponin content
(mg/g-1 Dry Wt.)
4
20
520
48
5.6
18
90
2,294
212
5.8
500
2,500
58,500
5,800
6.0
1000
50,000
108,000
120,000
33.5*
* Methyl-jasmonate (100 µM) treatment 7 days before harvest.
The pilot-scale 500 and 1000-L stainless bioreactor was designed according to the
BTBB type (Figure 2B). This reactor is comprised of a main body, air bubbling device,
steam generator for sterilization, air inlet, air vent system, and various control systems for
checking the temperature, oxygen, pH, and pipeline systems for transferring steam, air,
medium, and root masses (Figure 3). Additional equipments such as a distilled water
reservoir, medium mixer, medium sterilizer, and inoculation bioreactor are necessary.
167
www.taq.ir
Figure 2. Scale-up of hairy-like adventitious roots of Panax ginseng. (A) 20-L balloon-type
bubble bioreactors. (B) 500 and 1000-L pilot-scale balloon-type bubble bioreactors. (C)
10,000-L pilot-scale balloon-type bubble bioreactors for the commercial production of
ginseng roots. (D) Harvested ginseng roots from a 10,000-L pilot-scale balloon-type bubble
bioreactor. Photograph provided by Paek KY of CBN Biotech Co., Korea.
Figure 3. Schematic diagram of a balloon type bioreactor (A) and steam, air, and medium
flow (B) in pilot scale culture (1,000 L). 1, ventilation port; 2, light glass; 3, dissolved
oxygen probe port; 4, pH probe port; 5, inoculation port; 6, air inlet; 7, medium drain port;
8, stainless sparger; 9, sight glass; 10, screwed lid opener.
Before transfer to large-scale tanks, root tissues are homogenized into approximately
one cm length size and are moved via an air compressor though the inter-connector
between the inoculation reactor and the main tanks. The increase of the fresh weight of
ginseng roots was more than 30-fold after 40 days of culture in both bioreactors. The
168
www.taq.ir
biomass increase in this bioreactor was similar to the ginseng hairy root culture [43,69].
There is no serious problem with the stagnation of fluid flow and limit oxygen due to
the actively growing root mass. Based on the pilot-scale balloon-type bioreactor,
production of ginseng roots via 10,000-L bioreactor was practically attempted for the
commercial production (Figure 2C). In Korea, three companies produce the ginseng
roots commercially using pilot-scale bioreactor (10,000 to 20,000-L) and the basic
design follows the balloon-type bubble bioreactor. The root materials are processed into
various types of health foods and food ingredients (Figure 2D).
Acknowledgements
This work was funded in part by the Korea Research Foundation (F010608) and
Biogreen 21 of Rural Development Administration, Republic of Korea.
References
[1] Fujita, Y. (1988) Industrial production of shikonin and berberine. In: Applications of Plant Cell and Tissue
Culture, Ciba Foundation Symposium 137, Wiley, Chichester; pp. 228–238.
[2] Shimomura, K.; Sudo, H.; Saga, H. and Kamada, H. (1991) Shikonin production and secretion by hairy
root cultures of Lithospermum erythrorhizon. Plant Cell Rep. 10: 282–285.
[3] Srinivansan, V.; Pestchankar, L.; Moser, S.; Hirasuna, T.J.; Taticek, R.A. and Shuler, M.L. (1995) Taxolproduction in bioreactors: kinetics of biomass accumulation, nutrient uptake, and taxol production by cell
suspensions of Taxus baccata. Biotechnol. Bioengg. 47: 666-676.
[4] Kobayashi, Y.; Fukui, H. and Tabata, M. (1988) Berberine production by batch and semi-continuous
cultures of immobilized Thalictrum cells in an improved bioreactor. Plant Cell Rep. 7: 249-253.
[5] DiCosmo, F. and Misawa, M. (1995) Plant cell tissue culture: alternatives for metabolite production.
Biotechnol. Adv. 13: 425-453.
[6] Flores, H.E.; Hoy, M.W. and Puckard, J.J. (1987) Secondary metabolites from root cultures. Trends
[7] Ramachandra Rao, S. and Ravishankar, G.A. (2002) Plant cell cultures: Chemical factories of secondary
metabolites. Biotechnol. Adv. 20: 101-153.
[8] Chilton, M.D.; Tepfer, D.; Petit, A.; David, C.; Casse-Delbart, F. and Tempé, J. (1982) Agrobacterium
rhizogenes inserts T-DNA into the genomes of host plant root cells. Nature 295: 432–434.
[9] Lipp Joao, K.H.L. and Brown, T.A. (1994) Long-term stability of root cultures of tomato transformed
with Agrobacterium rhizogenes R1601. J. Exp. Bot. 45: 641-647.
[10] Baiza, A.M.; Quiroz-Moreno, A.; Ruiz, J.A. and Loyola-Vargas, V.M. (1999) Genetic stability of hairy
root cultures of Datura stramonium. Plant Cell Tissue Org. Cult. 59: 9-17.
[11] Flores, H.E. and Filner, P. (1985) Metabolic relationships of putrescine, GABA and alkaloids in cell and
root cultures of Solanaceae. In: Neumann, K.H.; Barz, W. and Reinhart, E.J. (Eds.) Primary and
Secondary Metabolism of Plant Cell Cultures. Springer-Verlag, New York; pp. 568–578.
[12] Kamada, H.; Okamura, N.; Satake, M.; Hirada, H. and Shimomura, K. (1986) Alkaloid production by
hairy root cultures in Atropa belladonna. Plant Cell Rep. 5: 239-242.
[13] Shanks, J.V. and Morgan, J. (1999) Plant ‘hairy root’ culture. Curr. Opin. Biotechnol. 10: 151-155.
[14] Giri, A. and Narasu, M.L. (2000) Transgenic hairy roots: recent trends and applications. Biotechnol.
Adv.18: 1-22.
[15] Hamill, J.D.; Robins, R.J.; Parr, A.J.; Evans, D.M.; Furze, J.M. and Rhodes, M.J. (1990) Overexpressing a yeast ornithine decarboxylase gene in transgenic roots of Nicotiana rustica can lead to
enhanced nicotine accumulation. Plant Mol. Biol. 15: 27-38.
[16] Hashimoto, T.; Yun, DZ. and Yamada, Y. (1993) Production of tropane alkaloids in genetically
engineered root cultures. Phytochem. 2: 713-718.
169
www.taq.ir
[17] Zhang, L.; Ding, R.; Chai, Y.; Bonfill, M.; Moyano, E.; Oksman-Caldentey, K.M.; Xu, T.; Pi, Y.; Wang,
Z.; Zhang, H.; Kai, G.; Liao, Z.; Sun, X. and Tang, K. (2004) Engineering tropane biosynthetic pathway
in Hyoscyamus niger hairy root cultures. Proc. Natl. Acad. Sci. USA 101: 6786-6791.
[18] Moyano, E.; Jouhikainen, K.; Tammela, P.; Palazon, J.; Cusido, R.M.; Pinol, M.T.; Teeri, T.H. and
Oksman-Caldentey, K.M. (2003) Effect of pmt gene overexpression on tropane alkaloid production in
transformed root cultures of Datura metel and Hyoscyamus muticus. J. Exp. Bot. 54: 203-211.
[19] White, F.F. and Sinkar, V.P. (1987) Molecular analysis of root induction by Agrobacterium rhizogenes.
In: Hohn, T. and Schell, J. (Eds.) Plant DNA infectious agents. Springer-Verlag, New York; pp. 149-177.
[20] Jung, G. and Tepfer, D. (1987) Use of genetic transformation by the Ri T-DNA of Agrobacterium
rhizogenes to stimulate biomass and tropane alkaloid production in Atropa belladonna and Calystegia
sepium roots grown in vitro. Plant Sci. 50: 145-151.
[21] Vernade, D.; Herrera-Estrella, A.; Wang, K. and Van Montagu, M. (1988) Glycine betaine allows
enhanced induction of the Agrobacterium tumefaciens vir genes by acetosyringone at low pH. J.
Bacteriol. 170: 5822–5829.
[22] Stachel, S.E.; Messens, E.; Van Montagu. M. and Zambryski, P. (1985) Identification of the signal
molecules produced by wounded plant cells that activate T-DNA transfer in Agrobacterium tumefaciens.
Nature 318: 624-629.
[23] Manickavasagam, M.; Ganapathi, A.; Anbazhagan, V.R.; Sudhakar, B.; Selvaraj, N., Vasudevan, A. and
Kasthurirengan, S. (2004) Agrobacterium-mediated genetic transformation and development of
herbicide-resistant sugarcane (Saccharum species hybrids) using axillary buds. Plant Cell Rep. 23:13443.
[24] Shimoda, N.; Toyoda-Yamamoto, A.; Nagamine, J.; Usami, S.; Katayama, M.; Sakagami, Y. and
Machida, Y. (1990) Control of expression of Agrobacterium vir genes by synergistic actions of phenolic
signal molecules and monosaccharides. Proc. Natl. Acad. Sci. USA 87: 6684–6688.
[25] Cangelosi, G.A.; Ankenbauer, R.G. and Nester, E.W. (1990) Sugars induce the Agrobacterium virulence
genes through a periplasmic binding protein and a transmembrane signal protein. Proc. Natl.Acad. Sci.
USA. 87: 6708-6712.
[26] White, F.F.; Taylor, B.H.; Huffman, G.A.; Gordon, M.P. and Nester, E.W. (1985) Molecular and genetic
analysis of the transferred DNA regions of the root-inducing plasmid of Agrobacterium rhizogenes. J.
Bacteriol. 164: 33-44.
[27] Taylor, B.H.; Amasino, R.M.; White, F.F.; Nester, E.W. and Gordon, M.P. (1985) T-DNA analysis of
plants regenerated from hairy root tumor. Mol. Gen. Genet. 201: 554–557.
[28] Brillanceau, M.H.; David, C. and Tempe, J. (1989) Genetic transformation of Catharanthus roseus G.
Don by Agrobacterium rhizogenes. Plant Cell Rep. 8: 63–66.
[29] Batra, J.; Dutta, A.; Singh, D.; Kumar, S. and Sen, J. (2004) Growth and terpenoid indole alkaloid
production in Catharanthus roseus hairy root clones in relation to left- and right-termini-linked Ri TDNA gene integration. Plant Cell Rep. 23: 148-154.
[30] Han, B.; Linden, J.C.; Gujarathi, N.P. and Wickramasinghe, S.R. (2004) Population balance approach to
modelling hairy root growth. Biotechnol. Prog. 20: 872-879.
[31] Bordonaro, J.L. and Curtis, W.R. (2000) Inhibitory role of root hairs on transport within root culture
bioreactors. Biotechnol. Bioeng. 70: 176-86.
[32] Shiao, T.L. and Doran, P.M. (2000) Root hairiness: effect on fluid flow and oxygen transfer in hairy root
cultures. J. Biotechnol. 83: 199-210.
[33] Weathers, P.J.; Wyslouzil, B.E.; Wobbe, K.K.; Kim, Y.J. and Yigit, E. (1999) The biological response of
hairy roots to O2 levels in bioreactors. In Vitro Cell. Dev. Biol.-Plant 35: 286-289.
[34] Doran, P.M. (1997) Hairy roots: Culture and Application. Harwood Academic Publishers.
[35] Hilton, M.G. and Rhodes, M.J. (1990) Growth and hyoscyamine production of 'hairy root' cultures of
Datura stramonium in a modified stirred tank reactor. Appl. Microbiol. Biotechnol. 33: 132-138.
[36] Kim, Y.H. and Yoo, Y.J. (1993) Development of a bioreactor for high density culture of hairy roots.
Biotechnol. Lett. 7: 859-862.
[37] Nuutila, A. M.; Lindqvist, A. S. and Kauppinen, V. (1994) Growth of hairy root cultures of strawberry
(Fragaria x ananassa Duch.) in three different types of bioreactors. Biotechnol. Techn. 11: 363-366.
[38] Kondo, O.; Honda, H.; Taya, M. and Kobayashi, T. (1989) Comparison of growth properties of carrot
hairy root in various bioreactors. Appl. Microbiol. Biotechnol. 32: 291-294.
[39] Uozumi, N.; Kohketsu, K. and Kobayashi, T. (1993) Growth and kinetic parameters of Ajuga hairy roots
in fed-batch culture on monosaccharide medium. J. Chem. Tech. Biotechnol. 57: 155-161.
[40] Doran, P.M. (1999) Design of mixing systems for plant cell suspensions in stirred reactors. Biotechnol.
Prog. 15: 319-335.
170
www.taq.ir
[41] Davioud, E.; Kan, C.; Hamon, J.; Tempé, J. and Husson, H-P. (1989) Production of indole alkaloids by
in vitro root cultures from Catharanthus trichophyllus. Phytochem. 28: 2675-2680.
[42] Keil, M.; Hartle, B.; Guillaume, A. and Psiorz, M. (2000) Production of amarogentin in root cultures of
Swertia chirata. Planta Med. 66: 452-457.
[43] Jeong, G.T.; Park, D.H.; Hwang, B. and Woo, J.C. (2003) Comparison of growth characteristics of
Panax ginseng hairy roots in various bioreactors. Appl. Biochem. Biotechnol. 107: 493-503.
[44] Kintzios, S.; Makri, O.; Pistola, E.; Matakiadis, T.; Ping, Shi H. and Economou, A. (2004) Scale-up
production of puerarin from hairy roots of Pueraria phaseoloides in an airlift bioreactor. Biotechnol.
Lett. 26: 1057-1059.
[45] Du, M.; Wu, X.J.; Ding, J.; Hu, Z.B.; White, K.N. and Branford-White, C.J. (2003) Astragaloside IV and
polysaccharide production by hairy roots of Astragalus membranaceus in bioreactors. Biotechnol. Lett.
25: 1853-1856.
[46] Jeong, G.T.; Park, D.H.; Hwang, B.; Park, K.; Kim, S.W. and Woo, J.C. (2002) Studies on mass
production of transformed Panax ginseng hairy roots in bioreactor. Appl. Biochem. Biotechnol. 98:
1115-1127.
[47] Kwok, K.H. and Doran, P.M. (1995) Kinetic and stoichiometric analysis of hairy roots in a segmented
bubble column reactor. Biotechnol. Prog. 11: 429-435.
[48] Tescione, L.D.; Ramakrishnan, D. and Curtis, W.R. (1997) The role of liquid mixing and gas-phase
dispersion in a submerged, sparged root reactor. Enzyme Microb. Technol. 20: 207-13.
[49] Cusido, R.M.; Palazon, J.; Pinol, M.T.; Bonfill, M. and Morales, C. (1999) Datura metel: in vitro
production of tropane alkaloids. Planta Med. 65: 144-148.
[50] McKelvey, S.A.; Gehrig, J.A.; Hollar, K.A. and Curtis, W.R. (1993) Growth of plant root cultures in
liquid- and gas-dispersed reactor environments. Biotechnol. Prog. 9: 317-322.
[51] Muranaka, T.; Kazuoka, T.; Ohkawa, H. and Yamada, Y. (1993) Characteristics of scopolaminereleasing hairy roots clones of Duboisia leichhardtii. Biosci. Biotech. Biochem. 57: 1398-13
[52] Kim, Y.J.; Weathers, P.J. and Wyslouzil, B.E. (2002) Growth of Artemisia annua hairy roots in liquidand gas-phase reactors. Biotechnol. Bioeng. 80: 454-464.
[53] Souret, F.F.; Kim, Y.; Wyslouzil, B.E.; Wobbe, K.K. and Weathers, P.J. (2003) Scale-up of Artemisia
annua L. hairy root cultures produces complex patterns of terpenoid gene expression. Biotechnol.
Bioeng. 83: 653-667.
[54] Kino-Oka, M.; Hitaka, Y.; Taya, M. and Tone, S. (1999) High-density culture of red beet hairy roots by
considering medium flow condition in a bioreactor. Chem. Eng. Sci. 54: 3179-3186.
[55] Williams, G.R. and Doran, P.M. (2000) Hairy root culture in a liquid-dispersed bioreactor:
characterization of spatial heterogeneity. Biotechnol. Prog. 16: 391-401.
[56] Dilorio, A. A.; Cheetam, R. D. and Weathers, P. J. (1992) Growth of transformed roots in a nutrient mist
bioreactor: Reactor performance and evaluation. Appl. Microbiol. Biotechnol. 37: 457-462.
[57] Whitney, P. J. (1992) Novel bioreactors for the growth of roots transformed by Agrobacterium
rhizogenes. Enzyme Mirobiol. Technol. 14: 13-17.
[58] Buer, C.S.; Correll, M.J.; Smith, T.C.; Towler, M.J.; Weathers, P.J.; Nadler, M.; Seaman, J. and Walcerz,
D. (1996) Development of a nontoxic acoustic window nutrient-mist bioreactor and relevant growth data.
In Vitro Cell. Dev. Biol.-Plant 32: 299-304.
[59] Wilson, P.D.G. (1997) The philot-scale cultivation of transformed roots. In: Doran, P.M. (Ed.) Hairy
roots: culture and application. Harwood Academic, Amsterdam; pp. 179-190.
[60] Taya, M.; Yoyama, A.; Kondo, O. and Kobayashi, T. (1989) Growth characteristics of plant hairy roots
and their cultures in bioreactors. J. Chem. Eng. Japan 22: 84-89.
[61] Holmes, P.; Li, S-L.; Green, K.D.; Ford-Lloyd, B.V. and Thomas, N.H. (1997) Drip-tube technology for
continuous culture of hairy roots with integrated alkaloid extraction. In: Doran, P.M. (Ed.) Hairy Roots:
Culture and Application; Harwood Academic, Amsterdam; pp. 201-208.
62] Bais, H.P.; Suresh, B.; Raghavarao, K.S.M.S. and Ravishankar, G.A. (2002) Performance of hairy root
cultures of Cichorium intybus L. in bioreactors of different configurations. In Vitro Cell. Dev. Biol.-Plant
38: 573-580.
[63] Kim, Y.; Wyslouzil, B.E. and Weathers, P.J. (2001) A comparative study of mist and bubble column
reactors in the in vitro production of artemisinin. Plant Cell Rep. 20: 451-455.
[64] Furuya, T.; Yoshikawa, T.; Orihara, Y. and Oda, H. (1994) Studies of the culture conditions for Panax
ginseng cells in jar fermentors. J. Natural Products 47: 70-75.
[65] Wu, J.Y. and Zhong, J.J. (1999) Production of ginseng and its bioactive components in plant cell culture:
current technological and applied aspects. J. Biotechnol. 68: 89-99.
171
www.taq.ir
[66] Asaka, I.; Li, I.; Hirotani, M.; Asada, Y. and Furuya, T. (1993) Production of ginsenoside saponins by
culturing ginseng (Panax ginseng) embryogenic tissues in bioreactors. Biotech. Lett. 15: 1259-1264.
[67] Choi, Y.E.; Jeong, J.H. and Shin, C.K. (2003) Hormone-independent embryogenic callus production
from ginseng cotyledons using high concentrations of NH4NO3 and progress towards bioreactor
production. Plant Cell Tissue Org. Cult. 72: 229-235.
[68] Hibino, K. and Ushiyama, K. (1998) Commercial production of ginseng by plant cell culture technology,
In: Fu, T.J.; Singh, W.R. and Curtis, W. (Eds.) Plant Cell Culture for the Production of Food Ingredients,
Proc ACS Symp, San Francisco, CA, USA, Plenum Press, New York; pp. 13-17.
[69] Yoshikawa, T. and Furuya, T. (1987) Saponin production by cultures of Panax ginseng transformed with
Agrobacterium rhizogenes. Plant Cell Rep. 6: 449-453.
[70] Kevers, C.; Jacques, Ph.; Thonart, Ph. and Gaspar, Th. (1999) In vitro root culture of Panax ginseng and
P. quinquefolium. Plant Growth Regul. 27: 173-178.
[71] Son, S.H.; Choi, S.M.; Soo, J.H.; Yun, S.R.; Choi, M.S.; Shin, E.M. and Hong, Y.P. (1999) Induction
and cultures of mountain ginseng adventitious roots and AFLP analysis for identifying mountain ginseng.
Biotechnol. Bioproc. Engi. 4: 119-23.
[72] Son, S.H.; Choi, S.M.; Lee, Y.H.; Choi, K.B.; Yun, S.R.; Kim, J.K.; Park, H.J.; Kwon, O.W.; Noh, E.W.;
Seon, J.H. and Park, Y.G. (2000) Large-scale growth and taxane production in cell cultures of Taxus
cuspidata (Japanese yew) using a novel bioreactor. Plant Cell Rep. 19: 628-633.
[73] Shin, K.S.; Murthy, H.N.; Ko, J.Y. and Paek, K.Y. (2002) Growth and betacyanin production by hairy
roots of Beta vulgaris in airlift bioreactor. Biotechnol. Lett. 24: 2067-2069.
[74] Yu, K.W.; Gao, W.Y.; Son, S.H. and Paek, K.Y. (2000) Improvement of ginsenoside production by
jasmonic acid and some other elicitors in hairy root culture of Ginseng (Panax ginseng C.A. Mayer). In
Vitro Cell. Dev. Biol. -Plant 36: 424-428.
[75] Yu, K.Y.; Gao, W.; Hahn, E.J. and Paek, K.Y. (2002) Jasmonic acid improves ginsenoside accumulation
in adventitious root culture of Panax ginseng C.A. Meyer. Biochem. Eng. J. 11: 211-215.
[76] Choi, S. M.; Son, S. H.; Yun, S. R.; Kwon, O. W.; Seon, J. H.; and Paek, K. Y. (2000) Pilot-scale culture
of adventitious roots of ginseng in a bioreactor system. Plant Cell Tissue Org. Cult. 62: 187-193.
172
www.taq.ir
OXYGEN TRANSPORT IN PLANT TISSUE CULTURE SYSTEMS
Oxygen transport limitations
WAYNE R. CURTIS1 AND AMALIE L. TUERK2
108 Fenske Laboratory, The Pennsylvania State University, University
Park PA-16802,USA - Fax:1-814- 865-7846 - Email: [email protected]
2
Department of Chemical Engineering, The Pennsylvania State
University, University Park, PA 16802
1
1. Introduction
The typical approach for teaching transport phenomena is from ‘first principles’ where
the physical model is simplified to point where it can be mathematically characterized.
The strength of this approach is that the mathematical description is rigorous – even
though the physical model may not be realistic. Often the rigorousness of the
mathematical description continues to be a sufficient means of characterizing the
system, even when the assumptions associated with the model are no longer valid. The
most common characterization of oxygen transport in gas-liquid systems is the lumped
parameter, kLa. The physical model for this situation is shown in Figure 1.
Figure 1. Simplified physical model of oxygen transfer based on well-mixed gas and liquid
phases. The resulting description of oxygen transfer rate OTR = kLa (DO*-DO) is widely
used to describe oxygen transfer in bioreactors.
173
© 2008 Springer.
www.taq.ir
W.R. Curtis and A.L. Tuerk
The liquid and gas phases within the bioreactor are lumped as effectively well-mixed
gas and liquid phases that are interconnected by the ‘limiting’ transport resistance
associated with the interfacial area per unit volume (a). The mathematical description
associated with this model is typical of mass transfer where the measure of conductance
(kLa) provides for transport in proportion to the concentration difference (“driving
force”), which is the deviation of the system from equilibrium.
OTR = kLa (DO* - DO)
(1)
In Equation 1, DO* is the equilibrium dissolved oxygen concentration in the medium,
which for aqueous systems at 25oC is roughly 258 µM or 8.24 ppm (exact values for
media depend on medium composition and atmospheric pressure [1]. DO is the bulk
liquid dissolved oxygen concentration. This simple equation has proven very useful for
characterizing oxygen transfer in a wide variety of bioreactors, including diffused air
systems where the assumptions of well mixed phases are clearly not valid. While this
limits the physical meaning of kLa (and prevents extrapolation to altered conditions),
the resulting logarithmic uptake of oxygen into a depleted liquid phase is behaviourally
valid for nearly any bioreactor configuration.
This paper presents an alternative approach for examining oxygen transport. The
starting point is the more realistic model of the bioreactor as a multiphase
heterogeneous system. The aim is not to develop rigorous mathematical descriptions,
but to understand the utility and limitations of commonly used mass transfer
relationships. This framework should provide a means of understanding oxygen
transport under conditions that cannot be readily characterized with mathematics.
Understanding what factors can be limitations to mass transfer is far more useful than
attempting to pragmatically guess at what should be the limiting factors for mass
transfer.
Figure 2. Schematic presentation of the different physical transport considerations for
oxygen transfer in a three-phase system. Transport is described in terms of transport within
a phase (INTRAphase) transport as well as between phases (INTERphase) transport. The
numerals and numbers correspond to the sections in which they are discussed. (e.g. II.A is
the section that examines gas-liquid interface mass transfer).
174
www.taq.ir
Oxygen transport in plant tissue culture systems
The framework for presenting oxygen transport is organized according to the physical
situations encountered in a multi-phase bioreactor system (Figure 2). Interphase oxygen
transfer refers to the transport of oxygen within a given phase, which includes the
fundamental mechanisms of diffusion and convection, as well as the less well-defined
concept of mixing. Interphase oxygen mass transfer refers to passing of oxygen from
one phase to another.
2. Intraphase transport
Oxygen transport within a phase should not be overlooked in bioreactor systems since it
is clearly the dominant form of transport. Nearly all of the oxygen that enters a
bioreactor, leaves the bioreactor in the gas phase without ever being transported to the
medium or tissue. In addition, while it is typical to focus on the gas and liquid phase,
oxygen consumption takes place within the plant tissue. The movement of oxygen from
the liquid in contact with the gas to liquid in contact with the tissue is of critical
importance. The transport of oxygen within these phases takes place by very different
mechanisms. Each of the three phases of gas, liquid and solid (tissue) are discussed
below.
2.1. OXYGEN TRANSPORT IN THE GAS PHASE
In most bioreactors, gas is the dispersed phase which is sparged into the system as gas
bubbles. While local mixing within a gas bubble is relatively rapid due to diffusion (and
small bubble size relative to mean free path), neither radial nor axial mixing of gas
within the reactor is assured. Since the general flow of gas is upward in a three-phase
system, axial mixing will only occur if there is sufficient axial liquid mixing to exceed
the rise velocity of the bubbles. For the low power levels used in agitation of plant cell
suspension culture [2], there will be minimal axial mixing. Radial mixing of a sparged
gas will occur to some extent as a result of rise-induced circulation cells. However, the
issue of dispersion of the gas bubbles does not really address the issue of mixing of the
gas phase. For mixing to occur, the bubbles must coalesce and breakup as they pass
through the vessel. Otherwise, each bubble acts as its own compartmentalized ‘batch’ of
gas, and only the residence time distribution of the gas will determine the extent of gas
transfer from the bubble. Measurements of gas-phase residence time distribution are
rather difficult and require techniques such as gas tracers and mass spectrometry [3].
“Fortunately”, the efficiency of oxygen transfer is so poor, that these issues of
dispersion and mixing within the gas phase are not typically very important because
there is not a large change in the gas phase composition as it passes through the reactor.
Even at extremely low gas flow rates, the composition of the gas exiting a vessel is
nearly the same as entering. While microbial reactors can be operated at sparge rates of
0.1-1 VVM (volumes of gas per volume of liquid per minute), a plant tissue culture
bioreactor can be operated at an order of lower magnitude gas flow rates and still have
minimal change in gas composition as a result of lower total respiration rates.
The gas phase can be the continuous phase within a bioreactor. This is true for a root
culture trickle-bed [4] or nutrient mist [5] bioreactors. In these systems, the gas flows as
a continuous stream from entrance to exit, and the liquid is dispersed (e.g. sprayed) and
175
www.taq.ir
passes through the reactor. Much like gas dispersed systems; the small change in gas
phase composition greatly simplifies the analysis. More importantly, the performance of
the bioreactor will not be dependent upon mixing with the bioreactor gas phase, and
assuming a constant well-mixed gas phase is a reasonable assumption. In the situation
of passive gas exchange in a plant tissue culture vessel (e.g. sponge plugs, plastic
closures or caps) the assumption of a uniform gas phase may be achieved; however, the
composition of the gas phase can be variable and unknown. We have measured
accumulation of carbon dioxide as much as 5% in a culture flask headspace – indicating
a significantly impaired exchange with ambient air (which is 0.03% CO2). Insufficient
gas exchange will reduce oxygen availability.
2.2. OXYGEN TRANSPORT IN THE LIQUID PHASE
Mixing in the liquid phase is highly dependent on bioreactor geometry and operational
conditions. For cell or tissue cultures that require more gentle conditions, the reduced
intensity of energy input will reduce liquid mixing. However, the time scale for growth
of plant tissues is very long relative to mixing times that would be encountered in most
liquid plant culture systems. It only takes a few seconds to completely mix a fluorescent
tracer in a shake flask culture [6]. However, mixing in a 15 L root culture took several
hours [3]. It is important to recognize the difference between mixing and circulation.
Both represent mechanisms of transporting oxygen throughout the bioreactor. Liquid
circulation is a measure of how fast a fluid element gets from one side of the bioreactor
to the other. Whereas, mixing is a measure of how quickly a fluid element can be
dispersed throughout the entire bioreactor. Achieving good liquid circulation can be
important to assure suspension of plant cell tissues. Liquid circulation can be greatly
affected by bioreactor geometry [7]. Note, however, that achieving greater bulk flow
throughout the reactor, does not necessarily imply better mixing. For example, low shear
paddle impellers which have proven effective in pilot scale plant cell suspension culture,
create flow, but lack the intense mixing of radial flow (Rushton) impellers [2]. Reduced
mixing should rarely be an issue for plant tissues because of their long culture times.
The bioreactor configurations used in plant tissue culture systems, are very varied as
compared to traditional fermentation. For example, fill and drain bioreactor
configurations (used in plantlet propagation) achieve liquid mixing as the media flows
in and out of the bioreactor [8]. In root cultures, the root matrix represents a tremendous
resistance to the flow and mixing of fluid [9]. In a gas-sparged (or air-lift) bioreactor,
liquid circulation and mixing results from flows induced by the differences in density
caused by the presence of the gas bubbles in the bioreactor. No matter what the specific
configuration, oxygen transfer to the plant tissues requires both mixing and circulation.
Mixing is required to disperse the oxygenated liquid in contact with gas to areas with
less oxygenation. Circulation is needed to move the oxygenated liquid to regions where
gas-liquid transport may not be as effective.
The extent, to which the liquid is mixed, has a fundamental impact on oxygen mass
transfer in larger vessels because the hydrostatic pressure (P) of the liquid in the tank
will increase oxygen transfer in the deeper parts of the tank. This is apparent from
Henry’s law, which describes the equilibrium oxygen solubility (CL*):
176
www.taq.ir
CL *
yO2 P
(2)
H
Developing equations which account for either the depth within the tank and the degree
of mixing within the liquid phase quickly becomes quite complex. Analytical solutions
are available for the limiting cases of complete axial mixing versus complete axial
segregation of the liquid phase [10]. Qualitatively the results can be understood in terms
of the impact of elevated oxygen transfer rates at the bottom of the bioreactor, and the
extent to which that liquid is circulated to other regions of the bioreactor.
Equation 2 is extremely important towards understanding various strategies of
enhancing oxygen transport in bioreactors. Most obvious is increasing the gas phase
oxygen mole fraction (yO2) through oxygen supplementation of the gas phase. The
effects of temperature are captured in the Henry’s law coefficient (H) where H increases
with temperature and the oxygen solubility is reduced. In this respect, the tendency to
grow plant tissues at 20-25oC is an advantage over E. coli or mammalian cell cultures
that have optimal growth rates at body temperature (37oC). By combining Equations 1
and 2, the complexity in rigorous description of oxygen mass transfer quickly becomes
apparent. The driving force for oxygen transfer throughout the reactor changes
depending on both depth and the composition of the gas phase. As mentioned in the
previous section, the analysis is simplified because the gas phase tends to remain
relatively constant within the vessel as a result of low rates of mass transfer relative to
the typical rates of gas introduction into the reactor.
A final condition worth noting for oxygen transport within the liquid phase is when
the culture medium has been solidified with agar or other gel matrix. Although the
medium is no longer a fluid, the gelled media is still 99% water and the rates of
diffusion of oxygen (and other nutrients) are indistinguishable from predictions based
on liquid diffusivities (unpublished data). For oxygen diffusion in water at 25oC, the
diffusion coefficient (DO2) is 2.26 x 10-5 cm2/s [11]. As will be discussed further below,
the diffusion rate of oxygen in stagnant water is also typically used to characterize
oxygen transfer rates within tissues.
2.3. OXYGEN TRANSPORT IN SOLID (TISSUE) PHASE
An organized tissue or cell aggregate can be oxygen deprived deep within the tissue
even if the surface is exposed to oxygen saturated medium. Cultured plants and plant
tissue present very large structures which must have considerable oxygen transport
within the tissue to maintain aerobic respiration. As the oxygen moves into the tissues,
it is consumed by respiration. The transport rate through the outermost tissues must be
sufficient to supply the oxygen to all tissues that are deeper within. A general (Cartesian
coordinate) mass balance for oxygen consumption within the tissue becomes:
wC O 2
wt
w
N O2 rO2
wx
(3)
177
www.taq.ir
The flux of oxygen (N O2) is described by Fick’s Law [e.g. N O2=Deff (wCO2/wx)], and rO2
is the biological oxygen demand (BOD) and associated conversion factors to obtain
consistent units (see Table 1). If the rate of oxygen consumption is dependent on the
tissue oxygen concentration, then solution of 3 is difficult. However, if the BOD is
assumed to be constant, the concentration profiles within the tissue are readily derived
from the steady state mass balance (wCO2/wt=0) based on the surface oxygen
concentration (CS). Table 1 presents these equations for various geometries that are
often used as approximation of tissues (plate, cylinder and sphere). The integration of
these equations assumes that there is no exhaustion of the oxygen within the tissue. The
assumption of ‘zero order’ oxygen use kinetics (BOD=constant) can be rationalized in
part because the tissues will invariably utilize any available oxygen before they would
resort to anaerobic respiration.
Table 1. Mass balance and oxygen concentration gradients within tissue that result from
diffusional mass transfer limitations.
Mass balance
Plate
Cylinder
Sphere
Concentration profile within tissue
w 2C
BOD U tissue
wx 2
1 §¨ BOD U tissue
2 ¨©
Deff
wC
wt
D eff
wC
wt
§ 1 · w § wC ·
D eff ¨ ¸ ¨ r
¸ BOD U tissue
© r ¹ wr © wr ¹
C
Cs wC
wt
§ 1 · w § wC ·
Deff ¨ 2 ¸ ¨ r 2
¸ BOD U tissue
© r ¹ wr © wr ¹
C
1 § BOD U tissue
Cs ¨
6 ¨©
Deff
C
Cs · 2
¸ L x2
¸
¹
1 §¨ BOD U tissue
4 ¨©
Deff
[4]
· 2
¸ R r2
¸
¹
· 2
¸ R r2
¸
¹
[5]
[6]
The diffusion of oxygen within the tissue phase is often assumed to be equivalent to
water (Deff=Do2,H20). The success of this approach is somewhat surprising given the
structural aspects of cells and convection associated with cytoplasmic streaming. It is
logical that an organism will transport oxygen throughout the tissue phase in such a way
that the net diffusion rate matches the oxygen transfer rate of the surrounding aqueous
system. Thus, the observation that the diffusion coefficient of oxygen in water is
comparable to the effective diffusion coefficient within a tissue (Deff) may reflect a
logical adaptation of the tissue physiology rather than a validation of diffusion as the
true transport mechanism. An example is provided on the use of these equations to
characterize oxygen transport in plant tissue culture in Section 4.
There are gas spaces within plant tissues-most notably within leaves. However, gas
spaces can develop in other tissues such as roots (aerenchema) under conditions where
they become oxygen deprived [12]. We have also observed hollow plant cell aggregates
that suggests the mechanism of tissue death to create these gas spaces is active even in
undifferentiated plant cells [2]. Although such structures clearly enhanced oxygen
transport, simple descriptions such as presented in Table 1 will not be useful. There are
also more complicated mechanisms of transport within differentiated plants in tissue
culture (e.g. Knudsen pore diffusion). The high humidity of a tissue culture vessel will
178
www.taq.ir
invariably limit transpirational convective flow and supply of sugar in the medium
(rather than synthesis in the leaves) will also alter ‘natural’ plant phloem transport. The
diversity of structures and tissues that are observed in plant tissue culture makes
generalizations difficult. An equally important determinant of transport gradients is the
rate of oxygen consumption. The impact of elevated BOD at tissue meristems is
discussed at the end of Section 3.
3. Interphase transport
3.1. OXYGEN TRANSPORT ACROSS THE GAS-LIQUID INTERFACE
The transport of oxygen across the gas-liquid interface is described in detail in all
biochemical engineering texts. Since gas phase diffusion is comparatively rapid, the
dominant resistance is in the liquid boundary layer. The subscript ‘L’ in kLa reflects this
observation, and a refined version of Equation 1 can be written to specify transport that
is taking place through the gas-liquid interface.
OTRg-L = kLa (CL* - CL)
(7)
The parameter ‘a’ is the interfacial area per unit volume. Because ‘a’ is not typically a
measurable quantity, the two parameters ‘kL’ and ‘a’ are lumped together as a single
parameter. The equilibrium dissolved oxygen level (CL*) is available for a wide variety
of conditions due to the fundamental importance for oxygen transport. There are
correlations for kLa that have been developed for a wide variety of bioreactor conditions
(e.g. agitator speed, gas sparge rate, reactor geometry); however, because the interfacial
area of a gas dispersion can be affected by so many operating conditions, application of
design equations to make predictions of OTR can be problematic. The example problem
in section 4 includes experimental determination of kLa and application to characterized
oxygen transport rates at the gas-liquid interface.
3.2. OXYGEN TRANSPORT ACROSS THE GAS-SOLID INTERFACE
Oxygen transfer at the gas-solid interface is rarely discussed in the context of biological
reactors. Similar to the situation of mass transfer from the gas to liquid, there is minimal
resistance to transport in the gas phase. As a result, oxygen delivery is limited by
transport within the tissue and the surface concentration (CS) will be determined by the
equilibrium relationship of Equation 2.
C S , g S
CL *
y O2 P
(8)
H
In contrast to microbial or other tissue culture systems, direct tissue-gas contact is
common in plant tissue culture. The ability of plant tissues to transport water and resist
desiccation, permits this type of growth for aseptic plants, callus and root culture. In
addition, intermittent liquid contacting [8] and even trickle-bed reactors [4] have
179
www.taq.ir
substantial tissue surface area that is exposed directly to gas. When a tissue is in contact
with gas, the characterization of oxygen transport is ‘simplified’ since the tissue surface
concentrations associated with internal oxygen transport is known and not calculated
iteratively with boundary layer mass transfer as is required for a solid-liquid interface
(Section 3.3).
3.3. OXYGEN TRANSPORT ACROSS THE SOLID-LIQUID INTERFACE
Mass transfer at a solid liquid interface is similar to the gas-liquid interface, only the
area of transport is more defined. As a result, the area is no longer lumped with the mass
transfer coefficient (kS) and the resulting equation is
OTRL-S = kS (Atissue / V)(CL - CS)
(9)
To obtain OTR per unit volume, the tissue surface area (Atissue) must be divided by the
culture volume (V). In this case, the ‘driving force’ for mass transfer is the difference
between the bulk dissolved oxygen level (CL) and the dissolved oxygen at the surface of
the tissue (CS). The mass transfer coefficient at a liquid-solid interface (kS) is dependent
on the extent of convection near the surface. There are hundreds of correlations that can
be used to estimate kS because they are generally used to describe mass and heat
transfer [13,14].
The scenario of a reaction being limited by transport at the fluid interface is a rather
challenging problem that is faced very frequently in non-biological and biochemical
reactors. As a result, there are many descriptions and approaches to solving this problem
in all reaction engineering texts and biochemical engineering texts. The general solution
to this problem is iterative: The net reaction depends upon the surface concentration and
the oxygen concentration profile that results from consumption and internal diffusion
(Equation 3). However, the net reaction also determines the required oxygen transfer
rate at the solid-liquid surface (Equation 9). The balance of boundary layer transport
and internal oxygen consumption can be found by choosing a surface concentration (CS)
then determining total internal oxygen consumption by integrating the internal
concentration profile (e.g. Table 1) and comparing oxygen transport at the surface until
it matches the boundary layer transport. If BOD can be considered independent of the
tissue oxygen level, then this approach is greatly simplified. Net reaction is calculated
directly from BOD, and the surface concentration is then fixed by the required boundary
layer transport rate. While the details of these approaches are not within the scope of
this chapter, these concepts are utilized in the analysis of example 4.3.
It is important to recognize that experimental measurements of tissue BOD are
unavoidably influenced by internal and external transport rates. As a result, the
measured oxygen consumption rates can actually be a combined measure of both tissue
oxygen consumption and solid-liquid mass transfer limitations. Correcting observed
BOD for the actual surface concentration was carried out in a recent evaluation of
respiration at the tips of hairy roots [15]. The key to carrying out assessments of oxygen
transport at the tissue-media interface is identifying an appropriate correlation for mass
transfer. These mass transfer correlations usually have the generalized form:
180
www.taq.ir
kS d p
DO 2
N Sh
§
f ¨ N Re
¨
©
U media v o d p
P media
, N Sc
P media
U media DO2
·
¸
¸
¹
(10)
where the equation is developed in terms of dimensionless groups: NSh is the Sherwood
number, NRe is the Reynolds number, and NSc is the Schmidt number. The major
determinant of the mass transfer coefficient is the extent of convection near the liquidsolid surface which is correlated within these equations as a bulk or superficial liquid
velocity (vo). Proper use of these correlations involves carefully matching units and
definitions used in the regression of the correlated experimental data.
A final important characteristic of plant tissues that affects liquid-solid transport
rates is growth from meristems. The high metabolic activity in a meristem results in
elevated meristematic BOD as compared to the bulk oxygen demand associated with the
majority of the tissue. Respiration in root culture meristems were measured as 10-times
greater than in the bulk [16]. A localized oxygen demand proportionately increases the
required mass transfer coefficients needed to avoid oxygen transport limitation.
Convection around this tissue must be much more intense than would be expected based
on assuming uniform distribution of total tissue BOD was assumed. When localized
meristematic oxygen demand is present, it must be accounted for by treating the high
BOD tissues separately from the bulk respiring tissue [4]. While the mathematical
treatment of localized meristem oxygen demand is rather involved, the important
qualitative implication of localized oxygen demand is that it greatly increases the
likelihood that the tissue respiration will be oxygen limited.
4. Example: oxygen transport during seed germination in aseptic liquid culture
The following section is presented to provide a specific application of the principles of
oxygen transfer. It also provides some experimental details on how this information can
be obtained and analyzed. Finally, the data presented should also clarify why oxygen
transport limitations are so common in cultured plant tissues, despite their apparent low
oxygen demand.
4.1. THE EXPERIMENTAL SYSTEM USED FOR ASEPTIC GERMINATION OF
SEEDS IN LIQUID CULTURE
The following experimental system provided a clear example of oxygen transport
limitation in plant tissue culture. The system was not created for this purpose; therefore,
the experimental system will only be described briefly with details being presented
elsewhere. Transgenic plants of Nicotiana benthamiana were created with a viral
replicase (REP) of bean-yellow dwarf geminivirus [17] expressed under the control of
the Aspergillus nidulans ethanol-inducible promoter [18]. Replicase gene insertion was
verified by PCR [(+)REP] and homozygous plants were generated by successive
‘selfing’ with selection based on the dominant kanamycin resistance gene. Seeds were
germinated in 50 mL of culture medium after surface sterilizing with 10% Clorox.
Germination took place in a Gamborg’s (B5) liquid medium [19] on a gyratory shaker
with 1.52 cm stroke at 150 rpm in a 25oC environmental incubator. Humidified air or
181
www.taq.ir
oxygen-enriched air were introduced into the shaker flask headspace at a flow rate of ~
15 mL/min after passing the gas through a 0.2 Pm gas sterilization filter.
4.2. EXPERIMENTAL OBSERVATION OF OXYGEN LIMITATION
Transgenic (+)REP seedlings germinated under ambient air conditions displayed
severely stunted hypocotyls (Figure 3).
Figure 3. Germination of transgenic N. benthamiana seeds that contain a viral replicase
(REP) under the control of an alcohol inducible promoter. WT = wild-type non-transgenic
seeds. (+)REP = homozygous plants. Inhibition of hypocotyl elongation results from
induction of REP as a result of alcohol formation during insufficient oxygen provided by
ambient oxygen (air). Error bars are standard deviation of ~30 seedlings.
Germination under 37% and 100% oxygen displayed a germination phenotype that was
indistinguishable from wild type plants. The lengths of hypocotyl segments were
measured by scanning the seedlings on a flat-bed scanner with a reference scale, then
digitizing length using the “NIH Image J image” analysis program. These results
suggest that under ambient air conditions, the germinating seeds experience sufficiently
anaerobic respiration to produce ethanol which induces the AlcA promoter and produce
the inhibitory replicase protein. Hypocotyl length for wild-type and (+)REP N.
benthamiana plants.
4.3. CHARACTERIZATION OF OXYGEN MASS TRANSFER
To provide a comparison of oxygen demand relative to oxygen transfer rates, the mass
transfer coefficient (kLa) was measured in shake flasks by adapting a sodium sulfite
oxidation test method [20]. The initial amount of Na2SO3 added to the flask
corresponded to the amount needed to react with an initially saturated water at 25oC (9.3
mg O2/L) plus a sufficient amount to react with 50% of iodine reaction indicator. The
method is based on the unreacted sulfite in a 10 mL sample reacting with 1mL of 0.025
N iodine under acidic conditions (0.5 mL glacial acetic acid). Then the unreacted iodine
is titrated with a 0.0025 N sodium thiosulfate solution (containing 1 g Sodium furoate
for stabilization) in conjunction with a saturated starch solution. KLa measurement was
182
www.taq.ir
carried out as replicated 2-point reaction rates (between 1 and 8 minutes) where the
reaction was initiated with 60 mL (50 mL water containing 6 Pg CoCl2 as the reaction
catalyst plus 10 mL containing 9.1 mg Na2SO3). The kLa measured for these
experimental conditions was 4.83 hr-1.
Carrying out measurements of oxygen uptake rate of germinating seeds as a function
of age is not within the scope of this report. Instead, it is known that respiration will
vary from essentially zero to values that are characteristic of meristematic tissue.
Meristematic tissues have considerably higher respiration rates [15,16]. The two basic
techniques used for BOD measurements are a submerged micro-dissolved oxygen cell,
and a Warburg respirometer [21]. In a dissolved oxygen cell, the BOD is calculated
based on the consumption of oxygen from the liquid phase: dC
BOD U tissue . The
dt
rate of oxygen usage is measured with a dissolved oxygen probe. The Warburg
respirometer measures the volume change in the gas phase as the carbon dioxide
evolved from respiration is absorbed into a basic solution [22]. It should be kept in mind
that both these techniques can only measure the rate of oxygen transport for the
experimental condition of the apparatus. As a result, the BOD values measured in this
way are directly impacted by mass transfer limitations such as the intra-tissue transport
and boundary layer transport described above. Correcting such observed values to
intrinsic BOD values is very involved [15]. For the purpose of this analysis, we have
chosen to use a range of BOD values of 0–100 Pmole/g fresh weight/hr based on
experience and reported literature values [1].
Mass transfer at the seed surface is estimated based on the rate of sedimentation of
the seeds. Although liquid mixing may be considerably faster than the seed
sedimentation rate, the seeds tend to move with the bulk flow; therefore, the
sedimentation rate provides a reasonable estimate of mass transfer at the surface. Seed
sedimentation velocities of 1.29 ± 0.059 cm/s (n=30) were measured in a glass tube.
Seed diameter estimated was 0.053 cm. The correlation for mass transfer coefficient
around a sphere is available as:
kS
ª
DO2 «
§ U media v S d p
2.0 0.6 ¨¨
«
dp «
P media
©
¬
1
· 2 § P media
¸ ¨
¸ Ü
¹ © media DO2
1
º
·3 »
¸
¸ »
¹ »
¼
(11)
Viscosity of water at 25oC is 0.89 cP. These conditions provide a seed surface mass
transfer coefficient of 0.00605 cm/s. The preceding analysis provides parameters
needed to examine oxygen transport for the seedling germination study. For the 40
seeds germinating in each flask, the total oxygen demand of the system would be 0.468
Pmoles per hour at a BOD of 100 Pmole / g FW /hr. If the BOD is considered a
constant, the minimum surface concentration of 216 PM can be calculated when the
center of the seed reaches a zero oxygen concentration from Equation 6:
Cs
BOD U tissue R 2
6 Deff
(12)
183
www.taq.ir
This shows that the dissolved oxygen level at the seed surface must approach the
ambient equilibrium dissolved oxygen (CL* = 250 PM) to avoid mass transfer
limitation. If the mass transfer limitation was only at the gas-liquid interface (Equation.
7, CL=CS), the total oxygen transfer capacity through the gas-liquid interface (V·OTRgL) would be 8.33 Pmoles per hour which is 18-times greater than the seed oxygen
demand. For the mass transfer limitations at the solid-liquid interface, the total oxygen
transfer to the 40 seeds can be calculated as 40(OTRL-s ·V) = kS(40·Aseed)(CL*-CS). This
provides a total transport rate at the media-seed interface of 0.265 Pmoles of oxygen per
hour, which is about half as much oxygen as the seeds require. These calculations
indicate that although the gas-liquid interface is not limiting oxygen transport, the
oxygen flux at the media-seed interface is insufficient to meet the oxygen demand.
Figure 4. Application of the oxygen transport equations to the example case study of seed
germination in a gyratory shake flask. Surface concentration of the seed (CS) is calculated
by Equation 12. Total biological oxygen demand (BOD) is compared to the total oxygen
that can be transported across the media-seed interface.
A more comprehensive analysis is presented in Figure 4. In this figure, the surface
concentration of the seed is calculated for the full range of BOD using Equation 6. The
remaining driving force (CL*-CS) is then used to calculate the transport at the seedmedia interface. [Note that to be totally rigorous, the bulk liquid concentration (CL)
would have to be corrected for the required gas-liquid transport; however, since that rate
is more than an order of magnitude higher than the solid-liquid interface, the correction
is very small for this example].
In this graph, oxygen deprivation is predicted within the germinating seed if the total
transfer rate for the seed-media interface is less than the total oxygen demand. As shown
in this figure, these calculations predict an oxygen limitation for germination under
ambient conditions. It should be kept in mind that the intention of these calculations is
not intended to be exact. It is very likely that the diffusion of oxygen within the
compact tissues of a seed will be considerably less than water. None-the-less, the
calculations are consistent with the observation of induction of the viral replicase as a
result of anaerobic metabolism. In addition, the calculations also predict that oxygen
deprivation can be prevented using an elevated oxygen partial pressure which is
consistent with the experimental observation of a wild-type phenotype for germinating
transgenic seeds at 37 and 100% oxygen.
184
www.taq.ir
5. Conclusions
The principles of oxygen mass transfer are presented to provide a qualitative
understanding of the culture conditions where oxygen transport limitations can be
observed. The context of the discussion is the applications of these principles to plant
tissue culture propagation vessels and bioreactors. An experimental system which
effectively uses an inhibitory protein driven by alcohol-inducible promoter is used as a
qualitative probe of oxygen deprivation in the germinating seeds. Oxygen limitation is
correctly predicted in this system even when the consumption rates of the seeds are
extremely small as compared to the gas-liquid oxygen transfer rates. It is shown that the
solid-liquid boundary layer is far more constraining for the delivery of oxygen. Use of
oxygen enrichment of the gas phase overcomes this mass transfer limitation by
increasing the driving force for transport in the bulk liquid phase. These principles of
oxygen mass transfer can be adapted (both qualitatively and quantitatively) to many
other aspects of oxygen-limited growth of plant tissues in culture.
Acknowledgements
Viral replicase construct with alcohol-inducible promoter was obtained from Hugh
Mason (Dept. Plant Biology, Arizona State University). Generation of the transgenic
plants was carried out through efforts of Jennifer Campbell, Jennifer Stick, Gregory
Thurber, Jason Collens, and Kelly Tender. Measurements of kLa were carried out with
the assistance of Randhir Shetty. Lauren Andrews carried out seed sedimentation
studies. Tobacco seeds were obtained from the <http://www.ars-grin.gov> USDA
National Plant germplasm system. Finally, we acknowledge financial support of the
National Science Foundation (REU supplement to Grant # BCS-0003926 & GOALI
program and) for A.L.T. and a Research Experience for Undergraduate site program
(Grant # EEC-0353569) for L.A.
References
[1] Curtis, W.R. (2005) Application of bioreactor design principles to plant micropropagation. Invited
contribution, 1st Int. Symp. on Liquid Systems for In Vitro Mass Propagation of Plants. Kluwer
Academic Publishers, The Netherlands; (in press).
[2] Singh, G. and Curtis, W.R. (1994) Reactor design for plant cell suspension culture. In: Shargool, P.D. and
Ngo, T.T. (Eds.) Biotechnological Applications of Plant Culture. CRC Press, Boca Raton, FL; pp.153184.
[3] Tescione, L., Ramakrishnan, D. and Curtis, W.R. (1997) The role of liquid mixing and gas-phase
dispersion in a submerged, sparged root reactor. Enz. Microbial Technol. 20: 207-213.
[4] Ramakrishnan, D. and Curtis, W.R. (2004) Trickle-bed root culture bioreactor design and scale-up:
Growth, fluid-dynamics, and oxygen mass transfer. Biotechnol. Bioeng. 88(2): 248-260.
[5] Kim, Y.J.; Weathers, P.J. and Wyslouzil, B.E. (2002) Growth of Artemisia annua hairy roots in liquidand gas-phase reactors. Biotechnol. Bioeng. 80(4): 454-464.
[6] Bordonaro, J.L. and Curtis, W.R. (1997) Development of a fluorescent tracer technique to evaluate mixing
in plant root culture. Biotechnol. Techniques 11(8): 597-600.
[7] Hsiao, T.Y.; Bacani, F.T.; Carvalho, E.B. and Curtis, W.R. (1999) Development of a low capital
investment reactor system: Application for plant cell suspension culture. Biotechnol. Prog. 15(1): 114122.
185
www.taq.ir
[8] Buwalda, F.; Frenck, R.; Lobker, B.; Berg-De Vos, B. and Kim, K.S. (1995) EBB and flow cultivation of
Chrysanthemum cuttings in different growing media. Acta Hort. 401:193-200.
[9] Carvalho, E. and Curtis, W.R. (1998) Characterization of fluid-flow resistance in root cultures with a
convective flow tubular bioreactor. Biotechnol. Bioeng. 60(3): 375-384.
[10] Tescione, L.; Asplund P. and Curtis, W.R. (1999) Reactor design for root culture: Oxygen mass transfer
limitation. In: Fu, T.J.; Singh, G. and Curtis, W.R. (Eds.) Plant Cell and Tissue Culture for the Production
of Food Ingredients. Kluwer Academic/Plenum Publishers, New York; pp. 139-156.
[11] Wilke, C.R. and Chang, P. (1955) Correlation of diffusion coefficients in dilute solutions. AIChE J. 1(2):
264-270.
[12] Ramakrishnan, D. and Curtis, W.R. (1994) Fluid dynamic studies on plant root cultures for application to
bioreactor design. In: Furusaki, S. and Ryu, D.D.Y (Eds.) Studies in Plant Science, 4: Advances in Plant
Biotechnology. Elsevier, Amsterdam; pp. 281-305.
[13] Cussler, E.L. (1997) Diffusion: Mass transfer in fluid systems. 2nd Edition, Cambridge University Press.
[14] Bennett, C.O. and Myers, J.E. Momentum Heat and Mass Transfer. 3rd Ed., McGraw Hill, 1982.
[15] Asplund, T.A. and Curtis, W.R. (2001) Intrinsic oxygen use kinetics of transformed root culture.
Biotechnol. Prog. 17: 481-489.
[16] Ramakrishnan, D. and Curtis, W.R. (1995) Elevated meristematic respiration in plant root cultures:
implications to reactor design. J. Chem. Eng. Japan 28(4): 491-493.
[17] Mor, T.S.; Moon, Y.S.; Palmer, K.E. and Mason, H.S. (2003) Gemini-virus vectors for high-level
expression of foreign proteins in plant cells. Biotechnol. Bioeng. 81(4): 430-437.
[18] Felenbok, B. (1991) The ethanol utilization regulon of Aspergillus nidulans: the alcA-alcR system as a
tool for the expression of recombinant proteins. J. Biotechnol. 17:11-18.
[19] Gamborg, O.L.; Miller, R.A. and Ojima, K. (1968) Nutrient requirements of suspension of soybean root
cells. Exp. Cell Res. 50: 148-151.
[20] Ruchti, G.; Dunn, I.J.; Bourne, J.R. and Von Stockar, U. (1985) Practical guidelines for the
determination of oxygen transfer coefficients (KLa) with the sulfite oxidation method. Chem. Eng. J.
30(1): 29-38.
[21] Carvalho, E.B. and Curtis, W.R. (2002) Effect of elicitation on growth, respiration and nutrient uptake of
root and cell suspension cultures of Hyoscyamus muticus. Biotechnol. Progress 18: 282-289.
[22] Umbreit, W.H.; Burris, R.H. and Stauffer, J.F. (1972) Manometric and biochemical methods applicable
to the study of tissue metabolism. Burgess Publishing Company, Minneapolis, MN.
186
www.taq.ir
TEMPORARY IMMERSION BIOREACTOR
Engineering considerations and applications in plant micropropagation
F. AFREEN
Department of Bioproduction Science, Chiba University, Matsudo, Chiba
271-8510, Japan-Fax: 81-47-308-8841-Email:[email protected]
1. Introduction
Commercial laboratories need to produce a large number of high quality plants at the
lowest possible costs of production which mainly includes labour cost, general overhead
cost and the cost per unit space in the growth room. Large-scale plant propagation by
using tissue culture technique is often criticized because of the intensive labour
requirement for the multiplication process; thus, scaling-up of the production systems
and automation of unit operations are necessary to cut down the production costs [1,2].
In order to achieve efficient and automated production in plant tissue culture, plant
production systems have evolved from a small research scale to a large volume and
high-yield culture system, and liquid media are preferably used to facilitate handling [3].
The use of bioreactors with liquid media for micropropagation is becoming more
popular due to the ease of scaling-up [4] and the low production costs [5]. Bioreactor is
a self-contained, sterile environment which capitalizes on liquid nutrient or liquid/air
inflow and outflow systems, and is mainly designed for intensive culture. The basic
function of a bioreactor is to provide optimum growth conditions by regulating various
chemical and/or physical factors. More specifically, it affords the maximal opportunity
to monitor and control over micro-environmental conditions such as agitation, aeration,
temperature and pH of the liquid medium. Several types of bioreactors are currently
available such as air lift-bioreactor, stirred tank bioreactor, rotating drum bioreactor,
column bioreactor etc. In these bioreactors, the plantlets or explants are cultured under
complete submerged condition in the liquid medium which may limit the gas exchange
of the plant materials and consequently result in vitrification or hyperhydricity of plant
tissues [6]. Vitrification is a severe physiological disorder involving apoplastic water
accumulation, due to the extended contact between the explants [7,8]. Symptoms of
vitrification include chlorophyll deficiency, cell hyperhydricity, hypolignification,
reduced deposition of epicuticular waxes and changes in enzymatic activity and protein
synthesis [7,8]. To avoid the problems associated with liquid culture in bioreactor,
different systems have been developed, such as membrane raft system, nutrient mist
bioreactor, temporary immersion bioreactor etc. [9]. Among those, temporary immersion
187
© 2008 Springer.
www.taq.ir
F. Afreen
bioreactor has gained popularity mainly due to its simplicity and high production rate
with minimum physiological disorders. In the current chapter the definition, brief
historical description, designing, benefits and related problems of the system will be
provided with special reference to the development of a new scaled-up system.
2. Requirement of aeration in bioreactor: mass oxygen transfer
Generally for normal plant cell metabolism, oxygen is required and only the dissolved
oxygen can be utilized by plants growing in an aqueous culture medium. Therefore, in a
bioreactor where oxygen transport limitations can usually be observed, aeration is
required to promote the mass transfer of oxygen from the gaseous phase to the liquid
phase. To meet the demand of the actively respiring plant tissues, forced-diffusion of
oxygen in the liquid nutrient medium is required and this can be achieved by aeration of
the liquid medium, agitation of the system, continuous shaking of the container etc.
Gas-liquid oxygen transfer can be explained by using the equation of Leathers et al. [3]:
OTR K La (C x C L )
(1)
Where, OTR is the volumetric oxygen transfer rate (mmol l-1 h-1), KL is the mass transfer
coefficient (m h-1), a is the specific gas-liquid interfacial area. The terms KL and a are
generally considered together and thus KLa in the current equation can be termed as
oxygen mass transfer coefficient (h-1). Cx is the dissolved oxygen concentration at
equilibrium with the gas phase (mmol l-1) and CL is the actual dissolved oxygen
concentration (mmol l-1) in the culture medium. KLa is frequently used to measure the
efficiency of oxygen transfer in a bioreactor. Oxygen solubility increases with
decreasing temperature; the dissolved oxygen concentration for 100% air saturated
water at sea level is 8.6 mg O2 /L at 25oC. The oxygen mass transfer coefficient is
strongly affected by agitation speed, air flow rate and design of a bioreactor. In general,
0.4
K La
P
· * V 0.5 * N 0.5
k §¨ 2
¸
s
s
V
R¹
©
(2)
Where, P2 is the power required to aerate the bioreactor, VR is the volume of the
bioreactor, Vs is the air flow rate, N is the agitation speed. Note that the mass transfer
coefficient increases with agitation speed and/or air flow rate. Most of the bioreactors
designed are capable to agitate (mixing) and aerate the medium simultaneously. In some
cases, such as in airlift bioreactor [10] to increase the dissolve oxygen concentration,
only aeration is used. In such case, N can be counted as zero. Many bioreactors have
been designed with liquid medium circulation system with the aim to improve the
oxygen transport. There are usually two different mechanisms of transporting oxygen
throughout the bioreactor, one is mixing and the other one is circulation. [see Curtis and
Tuerk in this volume]. As described by Curtis and Tuerk , liquid circulation is a
measure of how fast a fluid element gets from one side of the bioreactor to the other.
Whereas, mixing means, how quickly a fluid element can be dispersed throughout the
188
www.taq.ir
Temporary immersion bioreactor
entire bioreactor. However, achieving greater circulation throughout the bioreactor does
not necessarily result in better mixing. A detailed description of oxygen transport in
liquid culture system such as in bioreactor has already been described in this volume
[see Curtis and Tuerk]. In order to fulfil the oxygen demand of the cultured plants in the
bioreactor, a completely different approach has been taken, where, the plant materials
are exposed only temporarily to the liquid nutrient medium. Such a bioreactor does not
require any aeration or agitation and is termed as temporary immersion bioreactor.
3. Temporary immersion bioreactor
3.1. DEFINITION AND HISTORICAL OVERVIEW
The method of temporarily wetting the entire culture or plant tissue with nutrient
solution followed by the draining away of the excess nutrient solution under gravity so
that the plant tissue has access to air is defined as temporary immersion system. This
system usually involves a wetting and drying cycle which occurs periodically in a given
period of time and hence it can also be termed as periodic, temporary immersion. Heller
in 1965 [11], first mentioned that a mere up-and-down motion of the nutrient medium,
without renewal showed the same effect as a true renewal in suspension culture; this is
probably the first concept of the temporary immersion system. In 1985, Tisserat and
Vandercook [12], probably, first applied the idea of temporary immersion system in
plant tissue culture; they designed a system consisting of a large elevated culture
chamber that was drained and then refilled with fresh medium at certain intervals.
Aitken-Christie et al. in 1988 [13], developed a semi-automated culture system where
plant materials were cultured in a large container with automatic addition and removal
of liquid medium on a periodical base. After that, Simonton et al. [14] developed a
programmable micropropagation apparatus with cycled liquid medium; in this system
the liquid medium was intermittently applied to the cultured plants according to a
selected schedule. In order to overcome the physiological and technical limitations
encountered in bioreactors in the year 1993, a new temporary immersion system known
as RITA bioreactor was developed at CIRAD [15] This new technique has been used
for the improvement of plant propagation such as: banana [15], coffee [16], Hevea [17],
Citrus deliciosa [18] and many other plant species.
3.2. DESIGN OF A TEMPORARY IMMERSION BIOREACTOR
The principal components of a temporary immersion bioreactor are the same as those in
airlift or bubble column-type bioreactors, except, a fixed or floating raft support system
inside the culture vessel is required to support the explants. Liquid medium is pumped
into the culture vessel from a storage tank usually located underneath the vessel (Figure
1) or from a separate bottle in case of a twin bottle system.
189
www.taq.ir
F. Afreen
Figure 1. Design and operation procedure of a temporary immersion bioreactor.
The medium remains in the vessel for few minutes, after which it drains back to the
storage tank for reuse. The entire process is controlled by a solenoid valve and the
interval period varies from three to six hours depending on the plant species or
requirement of the explants.
3.3. ADVANTAGES OF TEMPORARY IMMERSION BIOREACTOR
Temporary immersion bioreactors provide an excellent way of using liquid medium at
the same time controlling the gaseous environment. Moreover, it can provide the
possible automation of the production system which facilitates low production costs. In
other words, increasing the rate of growth and multiplication by using bioreactors more
plants per unit area of the growth room are produced, which reduces the cost per plant
per unit space of growth room. Liquid culture bioreactors are mainly suitable for the
large-scale production of small size somatic embryos, growth of bulb, corms,
microtubers, compact shoot cultures etc.
Major features of a temporary immersion bioreactor are:
190
www.taq.ir
x
x
x
x
x
Reduction of hyperhydricity, compared with that of permanent immersion, is
the major achievement of a temporary immersion system. As plants are
immersed in the liquid medium only for 5-10 min. in every 3 or 6 h, the
physiological disorders are reduced and the plants become healthier.
Plant growth and development can be controlled by manipulating the
frequency and duration of immersion in liquid medium.
Plant growth is improved because during every immersion the plant is in direct
contact with the medium and a thin film of liquid covers the plant throughout
the interval period.
Air vents attached to the vessel prevent the cultures from contamination.
Due to the lack of agitation or aeration, the mechanical stress on plant tissues
are generally low compared with the other bioreactor systems.
3.4. SCALING UP OF THE SYSTEM: TEMPORARY ROOT ZONE IMMERSION
BIOREACTOR
The major problem imposed by liquid media in bioreactors even temporary immersion
bioreactor is the phenomenon of hyperhydricity, morphogenic shoot and leaf
malformation, due to the continuous immersion of the tissues in the medium [19]. The
malformations are manifested in glossy hyperhydrous leaves, distorted root and shoot
anatomy. Another important issue is the expression of contamination because sugarcontaining liquid medium in general encourages contamination. Exogenous
contamination can often be controlled by good sterile technique; however, endogenous
contamination cannot easily be controlled in repeated subcultures. To deal with these
problems, Afreen et al. [20] developed a scaled-up bioreactor known as temporary root
zone immersion bioreactor. The system is basically based on photoautotrophic (sugarfree medium) micropropagation and thus can reduce the chance of microbial
contamination. Moreover, the system can enhance the growth as well as improve the
quality of plants.
3.5. DESIGN OF THE TEMPORARY ROOT ZONE IMMERSION BIOREACTOR
The temporary root zone immersion bioreactor consisted mainly of two chambers
(Figure 2); the lower chamber was used as a reservoir for the nutrient solution and the
upper one for culturing embryos. A narrow air distribution chamber was located
between these two chambers. Two air-inlet tubes (internal diameter 5 mm; length 10
mm) opened into the air distribution chamber and were directly connected to an air
pump (Non noise S200, Artem Co. Ltd., Japan) via a filter disc (pore diameter 0.45 µm;
diameter 45 mm; Nippon Millipore Co. Ltd., Yonezawa, Japan) to prevent microbes
entering the culture vessel.
191
www.taq.ir
F. Afreen
Figure 2. Schematic diagram of the temporary root zone immersion (TRI-bioreactor)
bioreactor with forced ventilation system. Reproduced from Afreen et al. (2002) [20]).
The top of the air distribution chamber had several narrow tubes which were fitted
vertically in between the rows of the cell tray and opened in the culture chamber
headspace. The CO2 enriched air entered the culture chamber from the air distribution
chamber by means of these vertical tubes. Outflow was through four Millipore
membranes (pore diameter 0.45 µm; Nippon Millipore Co. Ltd., Yonezawa, Japan)
attached covering the outlet holes (10 mm diameter) on the sidewalls of the bioreactor.
The culture chamber contained a 6 cell by 9 cell autoclavable cell tray (Minoru Sangyo
Co. Ltd, Japan) for culturing the explants.
The nutrient reservoir chamber had an air inlet tube (a), which connected an air pump to
the headspace of the nutrient reservoir; an electric timer operated the pump. A second
tube (b) ran from close to the base of the reservoir to the culture chamber. To supply
nutrient solution to the culture chamber the air pump was switched on, thereby raising
the pressure in the headspace of the reservoir and forcing the nutrient solution from the
reservoir into the culture chamber. The nutrient solution immersed the root zone
temporarily for a total of 15 min every 6 h. After 15 min the air pump was switched off
and the excess nutrient solution flowed back into the reservoir under gravitation.
192
www.taq.ir
3.6. CASE STUDY – PHOTOAUTOTROPHIC MICROPROPAGATION OF COFFEE
Coffee plays a major role in the economy of many African, American and Asian
countries. The coffee plant is an evergreen, woody perennial that belongs to the
Rubiaceae family. The commercially important two species, Coffea arabica and Coffea
canephora were combined in a new species named Coffea arabusta [21]. The in vitro
growth of C. arabusta microcuttings is very slow [22] and therefore for the mass clonal
multiplication somatic embryogenesis is considered to be an effective, alternative
method.
In the multi-stage somatic embryogenesis of C. arabusta, cotyledonary stage is the
earliest stage embryo, capable of photosynthesizing [23]. However, the extent of
plantlet heterotrophy, photomixotrophy or photoautotrophy is dependent not only on
photosynthetic ability of the plant material but also on medium composition, volume of
culture vessels, aeration of the vessel etc. Therefore, Afreen et al. [20] cultured
cotyledonary stage coffee somatic embryos under photoautotrophic conditions in
different culture systems with the aim of developing an optimized protocol for largescale embryo-to-plantlet conversion and culture system.
The establishment and high PPF pre-treatment of somatic embryos have been
described by Afreen et al. [23]. Pre-treated cotyledonary stage embryos were selected
and then cultured under photoautotrophic conditions (in sugar-free medium with CO2
enrichment in the culture headspace and high PPF) in three different types of culture
systems as followed:
x Magenta vessel
x Modified RITA-bioreactor with temporary immersion system (Figure 1) and
x Temporary root zone immersion system bioreactor (TRI-bioreactor; Figure 2).
A mixture of vermiculite and paper pulp (as described by Afreen et al. [24]) was used as
supporting medium in the Magenta vessels and in TRI-bioreactors. For modified RITAbioreactors, MS liquid nutrient solution was used and the immersion frequency was 5
min/6 h by connecting an air pump through an electric timer. The planting density for
all the treatments was 2.4 X 103 plantlets/m2 area of culture tray.
To provide natural ventilation in the Magenta vessels, two gas-permeable Millipore
filter membranes (pore diameter 0.45 µm) were attached on the hole (10 mm diameter)
of the lid of the vessels. RITA-bioreactors were modified by attaching three gaspermeable filter membranes with 0.45 µm pore diameter and covering the hole (10 mm
diameter) of the lid of each of these vessels. The number of air exchanges was 2.6 h-1 in
both Magenta vessels and modified RITA-bioreactors throughout the experiment
(measured according to Kozai et al. [25]).
In TRI-bioreactor, forced ventilation was introduced by using an air pump
connected to the headspace of the air distribution chamber (Figure 2); the flow rates
were initially 50 ml min-1 (number of air exchanges was 1.6 h-1) and were gradually
increased every 2 or 3 days to maintain the CO2 concentration in the culture headspace
in a range ca. 1000 µmol mol-1, the maximum flow rate was 200 ml min-1 on day 45
(number of air exchanges was 5.8 h-1).
For all the treatments, hormone free MS medium was used as a basal medium;
sucrose, vitamins and amino acids were subtracted from the formulation to ensure the
photoautotrophic conditions. Vessels were placed in a growth chamber with an enriched
193
www.taq.ir
F. Afreen
CO2 concentration (1000-1100 µmol mol-1) and with a PPF of 100 µmol m-2 s-1 during
the 16 h photoperiod; ambient relative humidity was 80-85% and the air temperature
was 23oC.
Experiments were conducted for 45 days and the harvesting included recording of
plantlet conversion percentage, fresh and dry mass of the plantlets and percentage of
rooting. For the chlorophyll fluorescence, chlorophyll contents and stomatal studies ten
replicates were taken from each treatment. CO2 concentration in the culture headspace
was measured throughout the culture period and the net photosynthetic rate was
calculated according to the method of Fujiwara et al. [26]. Plantlets were transplanted in
the greenhouse (average temperature 29+2oC; RH 60-70%) and on Day 7 the survival
percentage was recorded. After 30 days of transplanting, plants were harvested and
fresh and dry mass of the survived plants were recorded.
In terms of plantlet conversion percentage the difference was very distinct among
the treatments; in TRI-bioreactor almost 84% of the cotyledonary stage embryos
produced plantlets, whereas in Magenta vessel and in modified RITA-bioreactor the
conversion percentages were 53 and 20% respectively [24]. Taking into account of all
the parameters of growth and development within the three different types of culture
vessels, it is evident that embryos grown in the TRI-bioreactor produced more vigorous
shoots and normal roots than those grown in Magenta vessel. The growth of the
plantlets attained in modified RITA-bioreactor was intermediate between that of
plantlets grown in the TRI-bioreactor and Magenta vessel (Figure 3).
The leaf fresh and dry mass of the plantlets from TRI-bioreactor were significantly
higher than those of the plantlets grown in modified RITA-bioreactor and Magenta
vessel. The most noticeable difference was observed in case of root growth. In TRIbioreactor, 90% of plantlets developed roots, 3 and 1.6 times more than plantlets grown
in modified RITA-bioreactor and Magenta vessel, respectively. It should be mentioned
here that even the roots which developed in a few plantlets in modified RITA-bioreactor
remained very small and stunted. Plantlets cultured in Magenta vessel exhibited an
intermediate root growth pattern between those of TRI-bioreactor and modified RITAbioreactor.
In TRI-bioreactor, as the plantlets grew in the course of time, the CO2 concentration
in the culture headspace was controlled by increasing the air inflow rate and thus the
number of air exchanges [24]. Thus, despite the increase in biomass, CO2 concentrations
were nearly the same throughout the experimental period (approx. 1280 µmol mol–1).
194
www.taq.ir
A
B
C
D
E
F
G
H
I
J
K
L
M
N
O
Figure 3. A. Coffee somatic embryos regenerated from leaf discs after 14 weeks of culture
under low light (30 µmol m–2 s–1) followed by 2 weeks under high light (100 µmol m–2 s–1)
(x0·5). B–D, 45-d-old plantlets developed from cotyledonary stage embryos under
photoautotrophic conditions in a temporary root zone immersion (TRI) bioreactor (B, x0.2),
a Magenta vessel (C, x0.7) and a modified RITA-bioreactor (D, x0.2). E–G, Stomata from
the abaxial (lower) surface of the first true leaves of plantlets developed
photoautotrophically in TRI-bioreactor (E), Magenta vessel (F) and modified RITAbioreactor (G). H and I, Individual plantlets immediately before transplanting ex vitro
grown in a TRI-bioreactor (H) and a Magenta vessel (I). J–L, Root development of plantlets
grown in a TRI-bioreactor (J), a Magenta vessel (K) and a modified RITA-bioreactor (L).
M–O, On day 30 after transplanting, plantlets previously grown in a TRI-bioreactor (M), a
Magenta vessel (N) and a modified RITA-bioreactor (O). Reproduced from Afreen et al.
(2002) [20]).
In contrast, in Magenta vessels and in the modified RITA-bioreactor, the number of air
exchanges could not be controlled, and were thus 3.3 h–1 throughout the experimental
195
www.taq.ir
F. Afreen
period (under natural ventilation). In the modified RITA-bioreactor, the CO2
concentration in the headspace fell from 1278 µmol mol–1 on day 7 to 1266 µmol mol–1
on day 42 despite the low air exchange rate; possible reasons for this low consumption
of CO2 by plantlets include:
x due to the small size of chlorophyllous plant materials, total CO2 consumption
is low;
x total chlorophyll contents of the plantlets are lower than those of plantlets in
other treatments; and most importantly,
x as the chlorophyllous plant material remained moist almost all the time due to
complete immersion of plantlets and the high humidity in the culture
headspace, these plantlets were probably virtually unable to fix any CO2 from
the atmosphere for in vitro metabolism.
The highest net photosynthetic rate was observed in plantlets grown in the TRIbioreactor [20]. In general, chlorophyll a and b contents (606 and 241 µg g–1 fresh mass,
respectively) based on the fresh mass of leaves was highest in plantlets grown in the
TRI-bioreactor, which were, 2 and 1.6 times, respectively those of leaves of plantlets
grown in the modified RITA-bioreactor. In the case of Magenta vessels, chlorophyll a
and b contents of leaves were intermediate between those of plantlets grown in TRIand modified RITA-bioreactors.
The potential activity of PSII (IpMAX), as estimated in the dark, was nearly the same
in leaves of plantlets grown in the TRI-bioreactor (IpMAX = 0.89) and in Magenta vessels
(IpMAX = 0.83) in contrast, IpMAX was low in leaves of plantlets grown in the modified
RITA-bioreactor (0.76). Similarly, in case of actual photochemical efficiency of PSII
(Ip) an increase in the quantum yield for electron transport was noted in leaves of
plantlets grown in both the TRI-bioreactor (Ip reaching 0.35) and in Magenta vessel (Ip
= 0.32), whereas the value was comparatively lower (Ip = 0.25) in plantlets of modified
RITA-bioreactor than those of plantlets in the other two treatments [20].
Microscopy highlighted that stomatal density was highest in the leaves of plantlets
grown in the TRI-bioreactor (8.3 mm–2 leaf area) followed by those of plantlets from the
modified RITA-bioreactor (7.5 mm–2 leaf area) and lowest in leaves of plantlets grown
in Magenta vessels (5.9 mm–2 leaf area). The most noticeable feature was that in the
leaves of plantlets from modified RITA-bioreactor some stomata were open wide while
others were distorted or still morphologically immature. It is possible that these stomata
may not function properly [20].
The survival percentage ex vitro of the plants, which was recorded on Day 7
followed a similar pattern and was highest (98%) in the plantlets grown in TRIbioreactor followed by 61% and 30% survival of the plants from modified RITAbioreactor and Magenta vessels, respectively.
The research [20] provides clear evidence that, for the embryo-to-plantlet
development under photoautotrophic conditions, the use of Magenta vessels and
modified RITA-bioreactor is less effective at promoting shoot and root growth both in
and ex vitro compared with the TRI-bioreactor. Moreover, for large-scale production the
use of small vessel has many disadvantages. On the other hand, RITA bioreactor is
claimed to be suitable for embryo-to-plantlet development without handling the plant
material [20]; however at the end of each phase the culture medium needs to be
changed. In case of RITA-bioreactor, density of plant material is also a limiting factor.
196
www.taq.ir
In general, RITA bioreactors are used for the development of plantlets from
embryogenic cell suspension cultures using sugar-containing medium. Therefore when
modified RITA-bioreactor was used for embryo-to-plantlet development under
photoautotrophic conditions, the growth was substantially reduced compared to the
growth obtained in TRI-bioreactor. This is most likely to be because in the modified
RITA-bioreactor after every immersion of the plant material with nutrient solution, the
entire plant becomes wet and, the plants remain covered by a film of nutrient medium
by capillary attraction during the interval period (Figure 4a).
Figure 4. Comparison between the Operation procedures of a) modified RITA-bioreactor
[16] and b) TRI-bioreactor [20].
In addition to this, because the relative humidity inside the vessel is normally high (9599%), the plant material either is never completely dried out or it takes a long period to
dry out. Thus, a thin layer of nutrient medium surrounding the plant material acts as a
liquid boundary layer, which impedes the exchange of gases between the plant and the
surrounding environment and possibly prevents the CO2 fixation in the chlorophyllcontaining zones - clearly a key factor for the photoautotrophic growth of embryos. In
case of conventional photomixotrophic systems, the media contain sugar and therefore
the lack of air exchanges may not be as serious a consequence as it is for the plantlets,
which completely depend on CO2 in the atmosphere for their photoautotrophic growth.
Again, it is emphasized that the RITA-bioreactor system has not been developed for
culturing plantlets under photoautotrophic conditions. Moreover, in this study, the
RITA-bioreactor was modified by attaching three gas permeable filter membranes on
the lid, as was done for Magenta vessels. Thus, a completely different result can be
expected if the original RITA-bioreactor with sugar-containing nutrient solution was to
be used.
197
www.taq.ir
F. Afreen
In contrast, in case of TRI-bioreactor only the root zone is immersed and the plant
remains undisturbed (Figure 4b). Therefore the exchange of gases between the plant and
the surrounding environment is unimpeded because there is no liquid boundary layer
resistance. In this situation, the plant can easily photosynthesize and produce its own
carbohydrate. Therefore, the TRI-bioreactor grown plantlets, not only exhibited the best
growth, but they were physiologically normal, survived well and grew faster ex vitro.
As discussed by Gupta et al. [27], in the conventional system, for embryo-to-plantlet
development following steps are necessary:
x Embryo selection and transfer on the germination medium.
x Germinated and rooted plantlet selection and transfer to soil.
x Acclimatization.
Generally, in each of the above phases, cotyledonary, late cotyledonary or germinated
somatic embryos are selected individually, in most cases by hand under the stereo
microscope. The invention of machine vision [28] and image analysis [29] systems offer
great potential for classifying and sorting embryos but the use is still limited. These
selected embryos are then transferred onto gelled medium for germination.
After 6-10 weeks of germination, plantlets with epicotyl are selected by hand,
transferred to soil and incubated in a greenhouse with frequent misting for
acclimatization and growth. In somatic embryogenesis procedures aimed at mass
production, these methods are still very time consuming and involve high labour
costing. However, in case of TRI-bioreactor system, cotyledonary stage embryo
selection is necessary which is done by hand, but once the embryos are transferred to
the bioreactor, germination, root development and acclimatization take place in the
same bioreactor and without handling the plant material or changing the culture
medium. Another advantage of the new system is that by increasing the number of cells
in the culture cell tray the density limitations can be overcome.
3.7. ADVANTAGES OF THE SYSTEM
x
x
x
x
x
x
x
Healthy, quality transplants or plantlets can be produced and the problem of
hyperhydricity can be reduced.
Microbial contamination is a major challenge to use liquid medium in
bioreactor system; by growing the plants in photoautotrophic conditions
(sugar-free medium) in TRI-bioreactor, this can be overcome very easily.
Most importantly it is ideally suitable for growing a variety of sizes of plantlets
starting from cotyledonary stage somatic embryos (0.6-1 cm) to 6-7 cm height
plantlets, which is not possible in other temporary immersion systems.
Unlike other bioreactors including temporary immersion bioreactor, the shoot
part remains undisturbed and thus the plant growth is not hampered.
After every immersion, the draining off of the excess nutrient reduces the risk
of nutrient stagnant condition.
Planting density limitations encountered in other systems can be overcome by
increasing the number of cells in the culture cell tray.
Planting density per self area can be increased significantly without reducing
the dry mass.
198
www.taq.ir
x
x
x
Handling is simple; once the bioreactor is filled and underway, the plants do
not require any attention other than assuring that the nutrient solution supply
system is operating properly.
If necessary, the pH, nutrient composition etc. can be easily measured and
controlled even during the production period.
Labour cost can be reduced at least 50% as large culture vessel are used in this
system.
4. Conclusions
For the large scale plant propagation purposes, bioreactors with liquid culture medium
can offer the most useful technique with many advantages over the other systems with
solid medium. Most importantly, the system can be automated and thus labor cost can
be reduced significantly. However, the occurrence of hyperhydricity of the propagules
hinders the commercialization of the system. The scaled-up system (TRI-bioreactor)
described in this chapter can overcome this problem successfully. The vigorous growth
and the higher survival percentage observed in plants from the TRI-bioreactor are the
cumulative results of many environmental and physiological factors during the in vitro
culture period: for example, the relative humidity in TRI-bioreactor under forced
ventilation was lower (85-90%) than that in the modified RITA-bioreactor (95-99%) or
in Magenta vessels (95%). The advantages of growing plants in an environment with
reduced relative humidity are manifold such as development of functional stomata,
increased wax deposition all of which can, in turn, prevent water loss when transferred
ex vitro and thus increase the chance of survival and subsequent growth. Furthermore,
in TRI-bioreactor the environmental parameters are maintained in such a way that the
difference between the in and ex vitro conditions is minimum, as a consequence when
the plants are transferred ex vitro they are capable to photosynthesize normally and thus
can easily overcome the transition stress during the first week of ex vitro condition.
Another important aspect is the supply of CO2 enriched air; the enhanced growth of
plants could have been largely due to the greater carbohydrate production of the plants
due to the supply of CO2.
We hope that the photoautotrophic culture system discussed here might also provide
the basis of a useful model for the in vitro propagation by somatic embryogenesis and
organogenesis of other important plant species. Future prospects of using TRI-bioreactor
are enormous. By using TRI-bioreactor it will be possible to reduce production costs to
a level lower than conventional propagation methods, making the products commercially
feasible. Recently this bioreactor has been used for propagation and increment of
medicinal concentrations of various medicinally important plants such as St. Johns
wort, Scutellaria baicalensis, Chinese licorice etc. Optimized environmental parameters
of the bioreactor can significantly influence secondary metabolite production and may
contribute to the development of an optimized and large-scale phytochemical
production system in bioreactor.
199
www.taq.ir
F. Afreen
References
[1] Aitken-Christie, J. (1991) Automation. In: Debergh, P. C. and Zimmerman, R. H. (Eds.)
Micropropagation. Kluwer Academic Publishers, Dordrecht, The Netherlands; pp. 342-354.
[2] Vasil, I. K. (1991) Rationale for the scale-up and automation of plant propagation. In: Vasil, I. K. (Ed.)
Scale-Up and Automation in Plant Propagation. Cell culture and Somatic Cell Genetics of Plants, Vol. 8.
Academic Press, San Diego; pp. 1-12.
[3] Leathers, R. R.; Smith, M. A. L. and Aitken-Christie, J. (1995) Automation of the bioreactor process for
mass propagation and secondary metabolism. In: Aiken-Christie, J.; Kozai, T. and Smith, M. A. L. (Eds.)
Automation and Environmental Control in Plant Tissue Culture. Kluwer Academic Publishers,
[4] Preil, W. (1991) Application of bioreactors in plant propagation. In: Debergh, P. C. and Zimmerman, R.
H., (Eds.) Micropropagation. Kluwer Academic Publishers, Dordrecht , The Netherlands; pp. 425-445.
[5] Paek, K. Y.; Hahn, E. J. and On, S. H. (2001) Application of bioreactors for large-scale micropropagation
system of plants. In Vitro Cell. Dev. Biol.- Plant. 37: 149-157.
[6] Debergh, P. and Maene, L. (1984) Pathological and physiological problems related to the in vitro culture
of plants. Parasitica 40: 69-75.
[7] Ziv, M. (1991) Quality of micropropagated plants - vitrification. In Vitro Cell. Dev. Biol.- Plant 27: 6469.
[8] Ziv, M. (1991) Vitrification: morphological and physiological disorders of in vitro plants. In: Debergh,
P.C. and Zimmerman, R.H. Micropropagation. Kluwer Academic Publishers, Dordrecht, The
[9] Akita, M. and Takayama, S. (1994) Stimulation of potato (Solanum tuberosum L.) tuberization by
semicontinuous liquid medium surface level control. Plant Cell Rep. 13: 184-187.
[10] Zobayed, S. M. A.; Murch, S. J.; Rupasinghe, H. P. V.; de Boer J. G.; Glickman, B. W.; and Saxena, P.
K. (2004) Optimized system for biomass production, chemical characterization and evaluation of chemopreventive properties of Scutellaria baicalensis Georgi. Plant Sci. 167: 439-446.
[11] Heller, R. (1965) Some aspects of the inorganic Nutrition of plant tissue cultures. In: White, P.R. and
Grove, A.R. (Eds). Proceedings of an International Conference on Plant Tissue Culture. England. pp. 1-8.
[12] Tisserat, B. and Vandercook, C. E. (1985) Development of an automated plant culture system. Plant Cell
Tissue Org. Cult. 5: 107-117.
[13] Aitken-Christie, J.; Singh, A. P. and Davies, H. (1988) Multiplication of meristematic tissue: a new
tissue culture system for radiata pine. In: Hanover, J.W. and Keathley, D.E. (Eds.) Genetic Manipulation
of Woody Plants. Plenum Press, New York; pp. 413-432.
[14] Simonton, W.; Robacker C. and Krueger S. (1991) A programmable micropropagation apparatus using
cycled liquid medium. Plant Cell Tissue Org. Cult. 27: 211-218.
[15] Alvard, D.; Cote, F. and C. Teisson (1993) Comparison of methods of liquid medium culture for banana
propagation. Effects of temporary immersion of explants. Plant Cell Tissue Org. Cult. 32: 55-60.
[16] Berthouly, M.; Dufour, M.; Alvaro, D.; Carasco, C.; Alemanno, L. and Teisson, C. (1995) Coffee
micropropagation in liquid medium using temporary immersion technique’. In: 16éme Colloque, Paris, 2,
pp. 514-519.
[17] Etienne, H.; Lartaud, M.; Michaux-Ferriére, N.; Carron, M. P.; Berthouly, M. and Teisson, C. (1997)
Improvement of somatic embryogenesis in Hevea brasilensis (Mull. Arg.) using the temporary
immersion technique. In Vitro Cell. Dev. Biol. -Plant 33: 81-87.
[18] Cabasson, C.; Ollitrault, P.; CoÃ te, F.; Michaux-Ferrie¡re, N.; Dambier, D.; Dalnic, R. and Teisson, C.
(1995) Characteristics of citrus cell cultures during undifferentiated growth on sucrose and somatic
embryogenesis on galactose. Physiol. Plant. 93: 464-470.
[19] Ziv, M. (2002) Simple bioreactors for mass propagation of plants. 1st Int. Symp. Liquid Systems for In
Vitro Mass Propagation of Plants, Ås, Norway , May 29th – June 2nd.
[20] Afreen, F.; Zobayed, S. M. A and Kozai, T. (2002) Photoautotrophic culture of Coffea arabusta somatic
embryos: Development of a bioreactor for the large-scale plantlet conversion from cotyledonary
embryos. Ann. Bot. 9: 20-29.
[21] Capot J. (1972) L’amelioration du cafeier en Cote d’Ivoire - Les hybrides ‘Arabusta’ Cafe Cacao The,
16: 3-17.
[22] Dublin, P. (1980) Multiplication vegetative in vitro de l Arabusta. Café–Cacao–The. Vol. WWIV, 4:
281-290.
[23] Afreen, F.; Zobayed, S. M. A. and Kozai, T. (2002) Photoautotrophic culture of Coffea arabusta somatic
embryos: Photosynthetic ability and growth of different stage embryos. Ann. Bot. 9: 11-19.
200
www.taq.ir
[24] Afreen, F.; Zobayed, S. M. A.; Kubota, C.; Kozai, T. and Hasegawa, O. (2000) A combination of
vermiculite and paper pulp supporting material for the photoautotrophic micropropagation of sweet
potato. Plant Sci. 157: 225-231.
[25] Kozai, T.; Koyama, Y. and Watanabe, I. (1998) Multiplication of potato plantlets in vitro with sugar free
medium under high photosynthesis photon flux. Acta Hort. 230: 121-127.
[26] Fujiwara, K.; Kozai, T. and Watanabe, I. (1987) Fundamental studies on environments in plant tissue
culture vessels. (3) Measurement of carbon dioxide gas concentration in closed vessels containing tissue
cultured plantlets and estimates of net photosynthetic rates of plantlets. J. Agric. Meterol. 43: 21-30.
[27] Gupta, P. K.; Timmis R. and Carlson, W. C. (1993) In: Soh, W.Y.; Liu, J.R. and Komamine, A (Eds.)
Advances in Development Biology and Biotechnology of Higher Plants. The Korean Society of Plant
Tissue Culture, Korea; pp. 18-37.
[28] Harrell, R. C. and Cantliffe, D. J. (1991) In: Vasil, I.K. (Ed.) Scale-up and Automation in Plant
Propagation. Academic Press, New York; pp. 179-195.
[29] Cazzulino, D.; Pederson, H. and Chin, C. K. (1990) In: Vasil, I.K. (Ed.) Bioreactors and Image Analysis
for Scale-Up and Plant Propagation. Academic Press, New York; pp. 147-175.
201
www.taq.ir
DESIGN AND USE OF THE WAVE BIOREACTOR FOR PLANT CELL
CULTURE
REGINE EIBL AND DIETER EIBL
Department of Biotechnology, University of Applied Sciences Wädenswil,
P.O Box 335, CH-8820 Wädenswil, Switzerland - Fax: 41-1-78850 Email: [email protected]
1. Introduction
Typical bioreactors for plant cell and tissue cultures have been made of glass or
stainless steel for more than 40 years. In this area, stirred reactors, rotating drum
reactors, airlift reactors, bubble columns, fluidised bed reactors, packed bed reactors and
trickle bed reactors with culture volumes up to 75 m3 as well as their modifications are
the most commonly used bioreactor types in research and commercial production
processes.
Disposable bioreactors represent modern alternatives to such traditional cultivation
systems. These bioreactors consist of a sterile plastic chamber that is partially filled
with media (10% to 50%), inoculated with cells and discarded after harvest. The singleuse chamber eliminating any need for cleaning or sterilisation is made of FDAapproved biocompatible plastics such as polyethylene, polystyrene and polypropylene.
Usually, the disposable bioreactors are low cost, simple to operate and guarantee high
process security. It is suggested that their use could improve process efficiency and
results by reducing the time-to-market of new products.
The aim of this chapter is to critically outline the potential of the disposable Wave
Bioreactor (hereafter referred to as Wave) based on wave-induced agitation for
secondary metabolite production from suspension cultures, hairy roots and embryogenic
cultures. With respect to the types of disposable bioreactor reported in the literature,
their classification, application and characterisation, here we describe the features of
Wave as well as summarise the results of hydrodynamic studies (characterisation of fluid
flow, estimation of mixing time, distribution time, energy input) and investigations of
oxygen transport efficiency. This allows a comparison of the Wave to other commonly
used bioreactors in plant cell based biomass as well as secondary metabolite production.
203
© 2008 Springer.
www.taq.ir
R. Eibl and D. Eibl
2. Background
2.1. DISPOSABLE BIOREACTOR TYPES FOR IN VITRO PLANT CULTURES
Table 1 gives an overview of the most frequently cited disposable bioreactors and
disposable bioreactor facilities for plant cell and tissue cultures, their schematic
diagram, manufacturers and described application.
Table 1. Disposable bioreactors (DB) and disposable bioreactor facilities (DBF) for
plant cell and tissue cultures
Mechanically driven membrane bioreactor
miniPerm® (DB)
Max. culture volume: 15 mL
Culture type: Embryogenic cultures
Application: Biomass production
Manufacturer: Sartorius AG
http://www.sartorius.com
Pneumatically driven bag bioreactor
LifeReactor® (DBP)
Ebb and Flow BioReactor (DBF)
Max. culture volume: 5 L
Culture type: Organogenic cultures (bud or
merismatic clusters), embryogenic cultures
Application: Micropropagation, production of
secondary metabolites
Manufacturer: Osmotek LTD
http://www.osmotek.com
Mechanically driven bag reactor
MantaRay ®(DB)
Culture type: Plant cell cultures
Application: No references
Manufacturer: Wheaton Science Products INC
http://www.wheatonsci.com
204
www.taq.ir
www.taq.ir
Design and use of the wave bioreactor for plant cell culture
Table 1. Disposable bioreactors (DB) and disposable bioreactor facilities (DBF) for
plant cell and tissue cultures. (continued)
Mechanically driven bag reactor
Optima and OrbiCell (DBF)
Culture type: Plant cell
cultures
Application: No references
Manufacturer: Metabios INC
http://www.metabios.com
Wave (DB and DBF)
Culture type : Callus
cultures, suspension cultures,
embryogenic cultures, hairy
roots
Application: Mass
propagation, production of
secondary metabolites
Manufacturer: Wave Biotech
AG (Switzerland)
http://www.wavebiotech.ch
http://www.wavebiotech.net
Wave Biotech LLC (USA)
http://www.wavebiotech.com
A air inlet, C - cells, E - gas exchange, G - gas exhaust, H - harvest, M – medium
205
www.taq.ir
R. Eibl and D. Eibl
In contrast to disposable bioreactor facilities (self-contained systems), disposable
bioreactors require external equipment such as incubators to provide the proper physical
as well as the necessary chemical environment for cells (e.g. temperature, aeration, pH
etc.) and to ensure monitoring and control of key process parameters. As indicated in
Table 1, there are generally two main types of disposable bioreactors (bioreactor
facilities), the choice of which depends on methods employed for supply of air and
mechanical energy of mixing: membrane reactors (mechanically driven) and bag
reactors (pneumatically and mechanically driven) [1-12].
Membrane bioreactors have been developed for the production of small product
volumes since the middle of the 80 s. Today their manufacturers offer specified
production chambers or modules, which can be chosen to suit the cells and product.
Müller-Uri and Dietrich [8] successfully applied the mechanically driven bioreactor
miniPerm® (Sartorius AG, Germany) equipped with a dialysis membrane for mass
propagation of proembryogenic suspension culture of Digitalis lanata. The main
disadvantage of membrane reactors consisting of a cultivation or production chamber
and a medium storage chamber (nutrient module) is their small culture volume.
Therefore, either the application of multiple units is required or the use of this reactor
type is restricted to research and production of high value compounds.
Larger culture volumes are offered by bag reactors. Bag reactors include bioreactors
in which the cultivation chamber is manufactured from plastic film and is designed as a
bag. For pneumatically driven bag reactors, the bag with the internal equipment such as
air sparger is fixed by a clamp arrangement, brought to a specified range of temperature
and aerated. The first disposable bioreactor for plant cell and tissue cultures cited in the
literature is a pneumatically driven bag reactor, namely a plastic bubble column. This
so-called LifeReactor® (Osmotek LTD, Israel) has a volume capacity of 2 L and 5 L and
is suitable for plant micropropagation (organogenic cultures of potato, banana,
pineapple, fern and orchid etc.) as well as cultivation of somatic embryos [9-12]. In
addition, two LifeReactors® based on temporary immersion technique and named Ebb
and Flow BioReactor® were constructed by Osmotek LTD (Rehovot, Israel). As
illustrated in Figure 1, the energy input of the Wave (Wave Biotech AG, Switzerland
and Wave Biotech LLC, USA) is caused by rocking the platform which induces a wave
(wave induced motion) in the bag with the cells in the medium. In this way,
oxygenation and mixing with minimal shear forces result. The surface of the medium is
continuously renewed and bubble-free surface aeration takes place. Optima® and
OrbiCell® reactors (Metabios INC, Canada) are based on a similar working principle.
2.2. THE WAVE: TYPES AND SPECIFICATION
Table 2 shows frequently used Wave Bioreactors for process scale-up (R & D,
laboratory scale, GMP manufacturing) and their technical specification. All the systems
facilitate measurement and regulation of rocking angle, rocking rate, temperature,
aeration rate as well as CO2 rate. Optional monitoring and control of pH, dissolved
oxygen as well as weight and flow rates in perfusion mode are possible. These are
typical process parameters [3,13-15] for the cultivation of plant cell and tissue cultures.
With only a few exceptions [16,17], addition of CO2 is necessary because of its positive
influence on biomass growth and secondary metabolite production, as in the case of in
vitro production of taxanes. However, the equipment of the Wave with an integral or
206
www.taq.ir
external aeration pump for plant cell cultivation usually achieves similar results without
the addition of CO2.
Figure 1. Working principle of the Wave.
Combining the laboratory Wave with an appropriate on-line analysing technique such as
ANTRIS, developed by Sensorix AG (Switzerland) and shown in Figure 2 on the right,
enables improved process control and allows realization of feeding strategies [18].
Figure 2. BioWave® 20 SPS with ANTRIS for on-line measurement of metabolites.
207
www.taq.ir
R. Eibl and D. Eibl
Table 2. Wave Bioreactors and their specifications.
BioWave® 2 SPS
BioWave® 20 SPS
BioWave® 200
SPS
Dimension
433 x 330 x 210 mm
720 x 580 x 400 mm
1900 x 1100 x 1100 mm
Performance
2 x Wave Bag 1 L1
2 x Wave Bag 2 L1
1 x Wave Bag 10 L1
2 x Wave Bag 2 L1
2 x Wave Bag 10 L1
1 x Wave Bag 20 L1
2 x Wave Bag 100 L1
1 x Wave Bag 200 L1
Scale
(maximum
culture volume)
R&D
(1 L)
Laboratory scale
(10 L)
GMP manufacturing
(100 L)
Agitation
Rocking rate from 6 to
42 rpm
Angle from 5 to 10°
42 rpm
25 rpm
Temperature
Integral heater or place in
incubator
Integral heater or place
in incubator
Integral heater
Aeration
Separate aeration unit and
flow meter
Integral aeration pump
Integral aeration pump,
flow meter and load cell
Standard
instrumentation
Temperature2,3; agitation
speed2,3; air flow
rate2,3;angle2
Temperature2,3;
agitation speed2,3; air
flow rate2,3; angle2
Temperature2,3;
agitation speed2; air
flow rate2,3
Optional
instrumentation
O22,3; CO22,3; pH2,3;
weight2,3,4
Temperature2,3; O22,3;
CO22,3; pH2,3; weight2,3,4
Agitation speed2,3;O22,3;
CO22.3; pH2,3; weight2,3,4
1
working volume or culture volume (filling level) of 50%, 2 measurement, 3 control, 4 perfusion module with
load cell
208
www.taq.ir
3. Design and engineering aspects of the wave
3.1. BAG DESIGN
Bioreactor, which forms the external cell environment, greatly influences plant cell line
growth and product formation. Existing bioreactor design concepts are based on
observations that the biosynthetic potential of a cell culture is closely linked to the
physical characteristics of cultivated cells and varies with cell line as well as culture
type. Thus, bioreactor design has to consider the morphology of cells including
differences between suspension and more differentiated organ cultures like hairy roots
for optimal cultivation. The biosynthetic capabilities of these cultures are not greatly
affected by their growth environment as long as the organised nature of the culture
morphology is maintained [1,19-22]. In the case of the Wave, this means that specially
designed cultivation bags are advantageous for different cell culture types (Figure 3).
Figure 3. Specially designed Wave Bags for different plant cell and tissue cultures.
209
www.taq.ir
R. Eibl and D. Eibl
The standard Wave Bag has an inoculation and sampling port, an inlet air as well as an
exhaust air filters, on-line probe insertion ports for pH, dissolved oxygen etc. and is
suitable for suspension cultures allowing inoculation via standard ports. If the cells
grow in aggregates and high biomass amounts are formed, a Wave Bag (Figure 3a) with
enlarged port to prevent the port quickly becoming clogged and screw cap for
inoculation and sampling is to be preferred. The Wave Bag shown in Figure 3b contains
a floating membrane of polyethylene, ensuring a perfusion mode in which suspension
cells can be continuously cultivated over a number of weeks. For hairy root cultivations,
a wasted nylon mesh is integrated into the bag (Figure 3c). The mesh acts as an
immobilisation matrix in order to prevent firstly the collection of free–floating roots at
one or two points in the cultivation chamber and secondly highly localised biomass with
a core of material which has lost its root morphology as well as productivity.
For plant cell and tissue cultures which do not release their products into the culture
medium, the biomass harvest before downstream processing of the product is necessary.
Under these circumstances, the formed biomass is removed by gloved hands after
opening the bag, which also allows lyophilisation. The different Wave Bag types
available in sizes from 2 L up to 100 L total volume have varying bag geometries,
which result in changing mass and energy transfer situations.
3.2. HYDRODYNAMIC CHARACTERISATION
As already proved, fluid dynamics (in particular fluid flow and fluid mixing)
encountered in a bioreactor are important factors for cell growth and production of
secondary metabolites based on plant cells (suspension cultures, hairy roots,
embryogenic cultures) [23-29]. A number of hydrodynamic studies have been carried
out for stirred and column reactors [26,30-32], but studies relating to the Wave, which is
still a relatively new cultivation system, are limited [33-36].
However, recent studies allow the comparison of the Wave to other commonly used
bioreactors. Consequently, one aspect of the work we have carried out is the
hydrodynamic characterization of the Wave. Our investigations were focused on fluid
flow, mixing time, distribution time, energy input and identification of interactions
between these features. All experiments were performed with standard Wave Bags and
water.
A modified Reynolds number (Remod) can be used to describe the fluid flow in the
Wave. The Reynolds number, which is the ratio of inertial force to internal friction, is
generally governed by Eq. (1) where w is the fluid velocity, l is the characteristic length
of the system (bag), and Ȟ is the kinematic viscosity of the culture medium.
Re
w*l
(1)
Q
In order to determine Remod, the characteristic length can be assumed to be a rectangular
cross-section calculated from liquid level (h) and width of the Wave Bag (B)
preconditioned steady state (Figure 4a). The liquid level of the bag is a function of
working volume (culture volume) and the bag geometry (i) is given by ratio of (L) to
(B). It is possible to correct deviations of the bag shape from a rectangular cross-section
210
www.taq.ir
by experimental determination (CAD) of true length (U) under liquid surface (Ao). The
fluid velocity (w) is defined as the ratio of medium flow rate (volumetric flow rate) to
the hydraulic cross-section (Aq); the volumetric flow rate ( V ) depends on the bag, the
working volume, the rocking angle (ĳ) as well as the rocking rate (k) of the Wave.
Depending on the combination of these four parameters, the volumetric flow rate varies
and as a result different amounts of substances are exchanged over the rotation point
(Figure 4b). The influence of the bag and rocking angle on volumetric flow rate can be
determined by experimental observations and calculated by introducing a correction
factor (C) obtained with the aid of regression analysis. Correction factors (C) for Wave
Bag 20 L are listed in Table 3.
Figure 4. Assumptions used to estimate Remod in the Wave. a) Initial position: ĳ=0, b) Final
position: ĳ=maximum.
211
www.taq.ir
R. Eibl and D. Eibl
Table 3. Correction factor (C) for Wave Bag 20 L. Reproduced from Lisica, S. (2004) with
permission [37].
Rocking
angle [°]
Working volume [L]
2
4
6
8
10
2
0.5354
0.2892
0.2025
0.1602
0.1323
4
0.819
0.5612
0.4083
0.3138
0.2583
6
0.9882
0.7628
0.5797
0.4554
0.3747
8
1.000
0.894
0.7167
0.585
0.4815
10
1.000
0.9548
0.8193
0.7026
0.5787
A correction factor (D), which depends on the bag type (Table 4), describes the
correlation of the Wave`s Remod and Remod occurring in stirred bioreactors.
Table 4. Correction factor (D) for Wave Bag. Reproduced from Lisica, S. (2004) with
permission [37].
Wave Bag
Correction factor (D)
Wave Bag 2 L
0.0565
Wave Bag 10 L
0.0398
Wave Bag 20 L
0.312
Wave Bag 100 L
0.015
Wave Bag 200 L
0.0489
Applying the correction factors (C) and (D), Remod for the Wave can be calculated as:
Re mod
V *k *C * D
15 *Q * ( 2 * h B )
(2)
Remod for Wave Bag 2 L, 20 L, 100 L and 200 L working with a constant rocking rate of
18 rpm and a rocking angle of 8° are illustrated in Figure 5a. It can be seen that Remod
decreases with increased filling level in Wave Bags working with higher volume.
Increased filling level results in reduced headspace volume as demonstrated in
Figure 5b, so that, the linear development of the wave movement is no longer possible
after a certain point. When using bags with large headspace, these phenomena did not
occur. Figure 5c shows Remod of Wave Bag 2 L working with 50% culture volume in
dependency on rocking rate and rocking angle. Remod increases according to the increase
212
www.taq.ir
of rocking rate and rocking angle. For different Wave Bags we were able to determine
the zone of Remod crit and established that Remod crit values range between 200 and 1000
(Figure 5d)).
(a)
(b)
(c)
(d)
Figure 5. Determination of Remod values for different wave bags.
Mixing time ș95 (time required to achieve 95% homogeneity) is measured by injecting a
tracer. It directly depends on the rocking rate and indirectly depends on the rocking
angle in the Wave [33,37]. The relationship between mixing time, rocking rate, rocking
angle and filling level for Wave Bag 200 L is shown in Figure 6. With the smallest
possible energy input (low rocking angle and rocking rate) and assuming identical
process parameters, the filling level of the Wave Bag significantly influences mixing
time, resulting in mixing time differences of over 100%. For higher rocking rates as
well as rocking angles, filling level has no significant effect on mixing time. Mixing
times based on 40% and 50% filling level lie between 10 s and 1400 s [36,38] for
Newtonian fluids (Table 5) and reach satisfactory values for cell culture bioreactors.
Even when there are specific production conditions (low rocking rate, low rocking
angle and filling level or medium to maximum rocking rate, rocking angle as well as
maximum filling level), mixing times generated in the Wave are comparable to
commonly used stirred reactors. Clearly, the most ineffective mixing (high mixing
times) takes place at the smallest possible rocking rate, rocking angle and maximum
filling level. Mixing time can be reduced by increasing the rocking rate and/or the
213
www.taq.ir
R. Eibl and D. Eibl
rocking angle, which results in a more intensive wave movement, rapid as well as
effective mixing.
Figure 6. Mixing times in BioWave® 200 SPS working with wave bag 200 L (40% and 50%
filling level).
Table 5. Mixing times of different Wave Bags working with 40% and 50% filling level.
Wave Bag
Mixing time [s]
2L
9 - 264
20 L
40 - 1402
100 L
22 - 837
200 L
65 - 874
The mixing time is a function of Remod and depends on the type of Wave Bag as well as
the filling level (Figure 7). The increase in Remod over values between 1000 and 2000
does not further reduce the mixing time. From Table 5, it becomes clear that the most
ineffective mixing of all bags investigated is shown by Wave Bag 20 L, which attains
mixing times lower than 100 s with considerably higher turbulences (Remod > 1500)
than other bag types. The most effective mixing is obtained with Wave Bag 2 L.
214
www.taq.ir
Figure 7. Mixing time as a function of Remod in different wave bags using 50% filling level.
Investigations focused on residence time distribution [39] have demonstrated that a
continuously operating Wave can be described by the ideally mixed stirred tank model.
In these experiments, the displacement technique was employed using
BioWave® 20 SPS. Figure 8 compares the measured response in the Wave and the
calculated residence time distribution in an ideally mixed stirred tank. Both curves are
congruent.
Figure 8. Comparison of measured residence time distribution in BioWave® 20 SPS.
(Ĳ=2.6 h, filling level=50%, rocking rate=6 rpm, rocking angle=5.1°) and theoretical
residence time distribution in an ideally mixed stirred tank.
In order to consider engineering aspects of a bioreactor system extensively, the
hydrodynamic characterization must also include energy input. In the case of the Wave,
the mechanical energy produced by the rocking platform facilitates mixing and improves
mass as well as heat transfer. First energy input modelling approaches [37,40] have
generated three static models, an inertia model, a momentum transport model, a model
215
www.taq.ir
R. Eibl and D. Eibl
for transformation into thermal energy and a model for electric power. Currently, static
model 3 is the most exact if we assume real flow behaviour in the bag. It is based on
films taken to calculate the momentums. The film sequences (30 per second) analysed
by CAD software show the actual distribution of fluid during wave movement. Static
model 3 is also valid for turbulent flow. In general, the static models presuppose a static
behaviour of the fluid in the bag. This assumption imposes the condition of equilibrium
for the sum of all acting momentums. Observing a cross-section of a bag at different
angles and in final positions presents the scenario in which the fluid movement is
finished. It can be seen that, the fluid is distributed according to the angles on the other
side of the rotation point. By analytical as well as graphical determination of the point
of gravity of the bag and the liquid surface, the resulting momentums can be calculated.
The energy input of the Wave is analogous to the work required for the movement
between the angles – ĳmax and + ĳmax.
Figure 9. Courses of specific energy input as a function of rocking rate, rocking angle,
maximum and minimum filling level for wave bag 2 L.
Figure 9 shows the courses of specific energy input as function of the rocking rate,
rocking angle and maximum as well as minimum filling level for Wave Bag 2 L.
Minimum filling level, maximum rocking angle and rate cause the maximum possible
energy input, which is one decimal power higher than operation with maximum filling
level. Up to rocking rates of 20 rpm, the specific energy input of all the systems is
directly proportional to the rocking rate. By increasing the rocking rate, the energy input
increases and reaches a stationary value limited by the technical specification of the
rocking unit. As a consequence of increased filling level, rocking rate and rocking angle,
a phase shift of the wave towards rocking movement occurs. Thus, the energy input is
slightly reduced at maximum filling level, with rocking angle and rocking rates greater
than 20 rpm. The energy input values of Wave Bag 2 L range from 8 to 561W m-3.
Some authors have determined the specific power input P/V or dissipation rate, in
particular cumulative energy dissipation, as a product of the energy dissipation rate and
the exposure time in an attempt to quantify the shear effects in stirred bioreactors
working with plant cells. Unfortunately, the obtained critical values based on significant
cell damage of 20% can vary considerably depending on cell line, cell age and culture
216
www.taq.ir
maintenance conditions. A critical value of energy dissipation of 107 J m-3 (104 J kg-1)
has often been reported [26,41-44]. This value corresponds to a specific power input of
about 111W m-3 for stirred reactors. For more sensitive mammalian cells, Henzler [30]
proposes an optimal range between 30 and 50W m-3.
3.3. OXYGEN TRANSPORT EFFICIENCY
Surface aeration is used to supply the medium containing the cells with oxygen in Wave
bioreactors. Experiments to determine the volumetric oxygen transfer coefficient (kL*a)
within the Wave (dynamic gassing-in method) using model media provide results
identical with published values of other bioreactors suitable for cell cultures [38].
The maximum value of kL*a measured was around 4 hr-1 at aeration rates of
0.002 vvm and 0.004 vvm in the Wave, whereas the maximum volumetric oxygen
transfer coefficient arising at 0.25 vvm was 9.8 hr-1. We finally obtained a value of kL*a
reaching 11.2 hr-1 at an aeration rate of 0.5 vvm. Under comparable cultivation
conditions, the values reported above are similar to those achieved in 1 L Biostat stirred
reactor with membrane aeration (4.5 hr-1 to 6.4 hr-1) from Sartorius BBI Systems
GmbH, Germany [45], 1.5 L stirred reactors with surface aeration (1.01 hr-1 to 3.1 hr-1)
[46], 8 L reactor with eccentric motion stirrer from Chema Balcke Dürr
Verfahrenstechnik GmbH, Germany, (maximum 13 hr-1) [47] and 15 L jar fermentor
with stirrer and aeration tube (Model MSJ-15, Marubishi Lab. Equip. Co. Ltd., Japan)
(about 10 hr-1) [48]. Volumetric oxygen transfer coefficients obtained by Knevelman et
al. [33] for Wave bioreactors are a decimal power higher than the values reported by
Rhiel and Eibl [34], Singh [35] and Eibl et al. [38]. However, they confirm the direct
relation to rocking rate and rocking angle.
Higher oxygen transfer efficiency results from increased energy input which caused
by increased rocking rate, rocking angle and aeration rate. A decreased filling level
increases kL*a at constant parameters. Oxygen transfer coefficients exceeding 11 hr-1 are
theoretically achievable in Wave bioreactors operated at high rocking rates and rocking
angles as well as aeration rates over 0.5 vvm, with direct dissipation of air into the
medium or application of pure oxygen. However, Wave Bag modifications would be
required to achieve such results. In the closed Wave Bag, oxygen transfer is limited.
Depending on bag size and filling level, oxygen saturation reaches 35% to 50%. Higher
saturation requires additional aeration.
4. Cultivation of plant cell and tissue cultures in the wave
4.1. GENERAL INFORMATION
Because of the sensitivity of plant cells to hydrodynamic shear stress, it is essential to
minimize the shear forces which occur during mixing and aeration generally. Exposure
of plant cells to high shear forces can reduce cell viability, change morphology and/or
aggregation pattern, impair growth and alter the concentration as well as the profile of
secondary metabolites significantly [26,29,43,44,49-52]. Based on results presented in
section 3, it can be deduced that the Wave guarantees optimal hydrodynamic conditions
217
www.taq.ir
R. Eibl and D. Eibl
for a large number of cell lines through adjustment of bag size, filling level, rocking
angle and rocking rate. Further reduction of hydrodynamic shear stress can be achieved
by use of viscous additive-supplemented media [53], addition of Pluronic®F-68 [26] and
cell immobilisation [1,54-56]. Shear stress and cell damage resulting from bubble rising
and bubble bursting does not occur in Wave systems (see 3.3, surface aeration).
The major physical cultivation conditions summarised in Table 6 have to be
maintained. A temperature between 25 and 27°C is one important parameter measured
and controlled in Wave systems. The pH is measurable and controllable by CO2 should
the necessity arise. The oxygen requirements and resulting aeration rates for most plant
cell and tissue culture cell types are low [3,57]. Where growth and product formation
are enhanced by the introduction of light, periodic illumination of cultures is possible
with external tubes installed around the Wave.
There is also a need for long-term sterility as a practical consequence of plant cell
and tissue cultures with relatively low growth rates (0.24 d-1 to 1.1 d-1 or doubling times
of 0.6d to 5d). Our experience shows that sterile Wave Bags can be used in plant cell
culture cultivation processes for up to 4 months. Contaminations by the bioreactor itself
are highly unlikely (less than 1%).
Table 6. Major physical cultivation conditions for plant cell and tissue cultures.
Parameter
Range
Temperature
25 - 27°C
pH
5.2 - 5.8
Aeration
0.1 - 0.3 vvm
Light
0 - 3000 Lux, often periodic light conditions
(16 hr on, 8 hr off)
In the case of long-term cultivation processes with middle and high culture volumes, the
application of filter heaters to prevent moisture build-up on exhaust air filters or the
periodic exchange of exhaust air filters is required. Table 7 shows the results of selected
batch and fed batch cultivations carried out in BioWave® 20 SPS with Wave Bag 2 L.
218
www.taq.ir
Table 7. Results of cultivations in BioWave® 20 SPS.
Culture type /
Species
Hyoscyamus muticus*
Panax ginseng**
Hairy roots
21 g L-1 d-1
fresh weight
Fed batch (feeding
and exchange)
20.3 g L-1 d-1
fresh weight
Fed batch
(feeding)
2.3 g L-1 d-1
fresh weight
Fed batch (feeding
and exchange)
5.1 g L-1 d-1
fresh weight
5.2 mg g-1
dry weight
hyoscyamine
5 mg g-1
dry weight
hyoscyamine
28 mg L-1
dry weight
ginsenosides
146 mg L-1
dry weight
ginsenosides
Reactor mode
Fed batch (feeding)
Biomass
productivity
Secondary
metabolite content
(max.)
Culture type /
Species
Taxus baccata***
Nicotiana tabacum
Suspension
culture
Reactor mode
Fed batch
(feeding), free
cells
Biomass
productivity
not determined
Secondary
metabolite content
(max.)
10 mg L-1 dry
weight paclitaxel
5 mg L-1 dry
weight baccatin
III
Fed batch (feeding),
immobilised cells
Batch
12 g L-1 d-1
fresh weight
20.8 mg L-1 dry
weight paclitaxel
7.8 mg L-1 dry
weight baccatin III
22 g L-1 d-1
fresh weight
none
219
www.taq.ir
R. Eibl and D. Eibl
Table 7. Results of cultivations in BioWave® 20 SPS.(continued)
Culture type /
Species
Allium sativum
Embryogenic culture
Reactor mode
Fed batch (exchange)
Biomass productivity
2.8 g L-1 d-1 fresh weight
Secondary metabolite
content (max.)
0.124 mg g-1 dry weight alliin
*Clone KB5 from Kirsi Oksman-Caldentey, Helsinki, Finland; **Clone T12 from Anna Mallol, Barcelona,
Spain; ***from Salima Bentebibel, Barcelona, Spain
In the following sections, we describe these experiments and our observations as well as
discuss the results in term of engineering and design aspects. The statements directed to
energy input based on static model 3 [37] (Figure 9). The theoretical predictions are
quite good in explaining the observed data and visual effects.
4.2. CULTIVATION OF SUSPENSION CULTURES
In the Wave working with 1L culture volume (bag with screw cap, Figure 3a), a tobacco
cell line was cultivated to evaluate the optimal parameters for biomass growth and
investigate the influence of increased energy input. The suspension culture (Nicotiana
tabacum) used was established and maintained in shake flasks at 25°C and 100 rpm in a
shaker-incubator as described by Rothe [7].
The batch cultivations of tobacco cells in MS medium were carried out in shake
flask with inoculum (30 and 50 g L-1 fresh weight) in logarithmic growth phase for 17
and 21 days. At a constant rocking angle of 6° or 10°, the rocking rates ranged from 17
to 25 rpm. Based on existing standard operation procedures in our group, the sampling
of the illuminated suspension cells was realised every second day to estimate the
biomass increases in terms of fresh weight and dry weight, the conductivity, the cell
viability, the pH as well as the sucrose consumption. Total biomasses between 290 and
220
www.taq.ir
432 g L-1 fresh weight were harvested under flow conditions in transition zone and
turbulent flow (Figure 10).
Figure 10. Influence of energy input on time course of biomass fresh weight and dry weight
for tobacco in the wave (1 L culture volume).
As illustrated in Table 7 and Figure 10, we achieved the highest biomass productivities
using 50 g L-1 inoculum, 0.2 vvm and about 1.4 times higher energy inputs at rocking
rates between 17 and 25 rpm as well as the highest possible rocking angle. This can be
explained by the improved mass transfer as a result of increasing culture broth viscosity
during biomass growth. However, the increase in the rocking rate does not favour
energy input in the Wave operating at a rocking angle of 10° with the maximum filling
level of bag 2 L, because the energy input is constant for rocking rates between 17 and
20 rpm (45W m-3). At a rocking rate of 25 rpm, energy input drops (35W m-3). In other
words, a further increase in rocking rate does not damage the cells, but increases the
oxygen transfer efficiency. To deliver higher energy inputs, it would be necessary to
decrease the rocking angle or filling level (see Figure 9). We have made similar
observations in Wave experiments working with suspension cultures of Vitis vinifera.
Bentebibel [1] describes the successful cultivation of free and immobilised
(Ca2+ alginate beads) cells of Taxus baccata growing in modified Gamborg`s B5
medium. Here, the process gain was the production of paclitaxel as well as baccatin III
in Wave Bags (equipped with screw cap) with 0.4 L working volume. The cultivations
running in fed batch mode for 24 days represent two-stage processes with a growth and
a production phase. The production phase was introduced using production medium
included elicitors. The initial culture volume was 0.25 L inoculated with 40 g fresh
weight of cell suspension in the growth phase. The experiments were carried out at
221
www.taq.ir
R. Eibl and D. Eibl
0.3 vvm, a constant rocking angle of 6°. An increase in the rocking rate from 20 to
40 rpm was made step by step as fresh culture medium was fed in. This strategy resulted
in a constant energy input of about 190W m-3 and flow in transition zone from laminar
to turbulent during the whole cultivation. It was found that immobilised suspension
cells of Taxus baccata produce 2-fold and 1.5-fold greater amounts of paclitaxel and
baccatin III than free suspension cells cultivated under comparable conditions in
BioWave® 20 SPS (Table 7). The obtained values of paclitaxel (10 to 20.8 mg
dry weight L-1) lie in the range of highest paclitaxel values reported earlier [16,59-62].
4.3. CULTIVATION OF HAIRY ROOTS
Studies with two transformed root lines also demonstrate the suitability of the Wave
(bags with screw cap) for hairy root cultivations under laminar fluid flow conditions.
The transformed root line of Hyoscyamus muticus (clone KB5, light-culture), supplied
by Dr. Kirsi Marja-Oksman, VTT, Espoo, Finland, produces intracellular tropane
alkaloids such as scopolamine and hyoscyamine in Gamborg`s B5 medium without
phytohormones [63,64]. Palazón et al. [6] discuss the procedures to cultivate
ginsenosides producing root line of Panax ginseng (clone T12, dark-culture) in SH
medium using different types of laboratory reactors.
Biomass growth and secondary metabolite production with the hairy root clones
used were promoted in the Wave (Wave Bag 2 L) by energy input values between 30
and 50W m-3. These values are comparables to those we applied for the cultivation of
tobacco cells. It is also important to note that parts of the roots should grow alternately
in submerged and emerged conditions by changing the position of the rocker unit.
Therefore, it is recommended to start cultivation with a minimum filling volume of
200 mL and energy input values of about 50W m-3 (6°, 6 rpm). The feeding is coupled
with a decrease in energy input (Figure 9), which maintains root integrity. An increase
in rocking rate in accordance with the medium feeding was characterised by changes in
morphology of both hairy root clones. We observed the formation of ball-like structures
which show poor growth and changes in branching, colour and root hair development.
In cultivations with increased energy input without feeding, a wound-response
(production of callus-type tissue) and the loss of biosynthetic capacity was noticed. It is
presumed that increases in energy input induce shear rates which represent stressful
conditions for the growing roots, although not sufficient enough to disrupt them.
The data in Figure 11 for the Hyoscyamus muticus hairy roots indicate significant
increase in biomass using the BioWave® 20 SPS with the optimal culture conditions
[3,5]. Independent of bioreactor mode, the growth of root biomass containing tropane
alkaloids (5 mg g-1 dry weight hyoscyamine) was about 120-fold after 28 days
(Table 7). This is the highest biomass productivity of the laboratory cultivation systems
investigated. The biomass produced maintained their typical morphology.
Under optimum conditions, as described above, it has been reported [6] that Wave
cultured ginsenosides producing hairy roots can enhance root fresh weight more than 5fold compared to 3.7-fold in an emerged spray reactor. From periodic medium
exchanges and doubling the cultivation time, 28-fold higher biomass increases in the
Wave and 12.1-fold higher biomass increases in the spray reactor result. While the
maximum ginsenoside productivity has been reached 2.6 mg L-1 d-1 in the Wave, the
maximum ginsenoside productivity in the spray reactor was 0.7 mg L-1 d-1. The first run
222
www.taq.ir
of Wave Bag 20 L (maximum 5 L culture volume) provided 423.6 g ginsenosides
biomass (total 214 mg L-1 ginsenosides) in 52 days. Through the use of the special hairy
root bag with integrated mesh, the highly localised root mass loses its typical root
morphology and should be avoided at higher culture volumes of 0.5 L.
Figure 11. Biomass increase of hyoscyamine producing hairy roots (Hyoscyamus muticus,
clone KB5) for different cultivation systems.
Follow-up tests indicated the possibility of direct inoculation with hairy roots from
plates for both hairy root clones. No differences in hairy root morphology, biomass
growth and secondary metabolite production were detected in experiments with
inoculum from plates and shake flasks. The shake flask mass propagation procedure for
inoculum production can therefore be omitted. This results in reduced time and process
costs.
4.4. CULTIVATION OF EMBRYOGENIC CULTURES
Embryogenic culture of Allium sativum was established to produce alliin in laboratory
stirred reactors, column reactors and the Wave [57]. The cells were grown in modified
MS medium in shake flasks (100 rpm, dark). Provided with an inoculum of 30 g L-1
fresh weight, the cells were cultured in fed batch (medium exchange in production
phase) liquid suspension with light (25°C, 0.2 vvm). The Wave cultivations were
performed in Wave Bags 2 L with screw cap at a constant energy input of 70W m-3
corresponding to 0.5 L culture medium, 6° angle and 11 rpm for 28 days. In the Wave,
20% higher biomass production was achieved under fluid flow in the transition zone
from laminar to turbulent. The maximum biomass productivity was approximately
2.8 g L-1 d-1 fresh weight, yielding 0.124 mg g-1 dry weight alliin (Table 7).
223
www.taq.ir
R. Eibl and D. Eibl
5. Conclusions
For plant cell and tissue cultures, disposable bioreactors such as the Wave provide an
efficient alternative to standard glass or steel bioreactors. Its application in process
development as well as in small and middle volume commercial production processes
can increase process safety and reduce time as well as process costs. For example, timeintensive cleaning and sterilisation procedures as well as intermediate steps for
inoculum production can be omitted. Biomass as well as secondary metabolite
production in Wave bioreactors is comparable or even higher than in traditional
laboratory reactors. This is a consequence of optimum hydrodynamic characteristics for
hairy roots, suspension cultures and embryogenic cultures. High shear stress can be
countered by high filling volume, minimum rocking rates and angles. Because of these
characteristics and, in addition, its scale-up capability, the Wave has enormous potential
for efficient commercial production processes based on plant cells. We expect this
potential to be verified in the near future.
Acknowledgements
The authors` research was partly supported by the Commission of Technology and
Innovation in Switzerland (CTI).
References
[1] Bentebibel, S. (2003) Estudio de la producción de taxanos por cultivos de células en suspensión e
inmovilzadas de Taxus baccata. Ph D Thesis, University of Barcelona, Barcelona.
[2] Eibl, R. (2002) Fermentative Herstellung bioaktiver Wirkstoffe mit dem Wave. BioWorld 6: Sonderdruck
BioteCHnet.
[3] Eibl, R. and Eibl, D (2002) Bioreactors for plant cell and tissue cultures. In: Oksman-Caldentey, K.M. and
Barz, W. (Eds.) Plant Biotechnology and Transgenic Plants. Marcel Decker, Inc., New York; ISBN 08247-0794-X; pp. 163-199.
[4] Eibl R (2003) Pflanzliche Zell-und Gewebekulturen-Wirkstoffproduzenten mit Zukunftspotential.
Drogenreport 30: 17-19.
[5] Lettenbauer, C. and Eibl, R. (2001) Application of the wave bioreactor 20 for hairy root cultures. In:
Wildi, E. and Wink, M. (Eds.) Trends in Medicinal Plant Research: Screening, Biotechnology and
Rational Phytotherapy. Romneya-Verlag, Dosenheim; ISBN 3934502024; pp. 139-141.
[6] Palazón, J.; Mallol, A.; Eibl, R.; Lettenbauer, C.; Cusidó, R.M. and Piñol, M.T. (2003) Growth and
ginsenoside production in hairy root cultures of Panax ginseng using a novel bioreactor. Planta Med. 69:
344-349.
[7] Rothe, S. (2004) In vitro Produktion kosmetischer Wirkstoffe mit Pflanzenzell- und Gewebekulturen.
Diploma Thesis; University of Applied Sciences Giessen Friedberg, Giessen.
[8] Müller-Uri, F. and Dietrich, B. (1999) Kultivierung proembryogener Massen von Digitalis lanata im
MiniPerm Bioreaktor. In Vitro News 3: 4.
[9] Harrell, R.C.; Bieniek, M.; Hood, C.F.; Munilla, R. and Cantliffe, D.J. (1994) Automated in vitro harvest
of somatic embryos. Plant Cell Tissue Org. Cult. 39: 171-183.
[10] Fukui, H. and Tanaka, M. (1995) An envelope-shaped film culture vessel for plant suspension cultures
and metabolite production without agitation. Plant Cell Tissue Org. Cult. 41: 17-21.
[11] Ziv, M.; Ronen, G. and Raviv, M. (1998) Proliferation of meristematic clusters in disposable
presterilized plastic bioreactors for large-scale micropropagation of plants. In Vitro Cell Dev. Biol.-Plant
34: 152-158.
224
www.taq.ir
[12] Escalona, M.; Lorenzo, J.C.; Gonzalez, B.L.; Daquinta, M.; Gonzalez, J.L.; Desjardine, Y. and Borroto,
C.G. (1999) Pineapple (Ananas comosus L. Merr) micropropagation in temporary immersion systems.
Plant Cell Rep. 18: 743-748.
[13] Curtis, W.R. (1999) Achieving economic feasibility for moderate-value food and flavour additives: A
perspective on productivity and proposal for production technology cost reduction. In: Fu, D.J.; Singh, G.
and Curtis, W.R. (Eds.) Plant Cell and Tissue Culture for the Production of Food Ingredients. Kluwer
Academic Press, New York; ISBN 0-306-46100-5; pp. 225-236.
[14] Endress, R. (1994) Plant Cell Biotechnology. 1st Edition. Springer Verlag, Heidelberg; ISBN 3-54056947-2; pp. 46-83.
[15] Lee, B.H. (1996) Fundamentals of Food Biotechnology. 1st Edition, VCH Publishers, Inc., New York;
ISBN 1-56081-694-5; pp. 355-370.
[16] Mirjalili, N. and Linden, J.C. (1995) Gas phase composition effects on suspension cultures of Taxus
cuspidata. Biotechnol. Bioeng. 48: 123-132.
[17] Linden, J.C.; Haigh, J.R.; Mirjalili, N. and Phisaphalong, M. (2001) Gas concentration effects on
secondary metabolite production by plant cell cultures. Adv. Biochem. Eng. Biotechnol. 72: 28-62.
[18] Pechmann, G.; Ducommun, C.; Lisica, L.; Lisica, S.; Blum, P.; Eibl, R.; Eibl, D.; Schär, M.; Wolfram,
L.; Rhiel, M.; Emmerling, M.; Röll, M.; Lettenbauer, C.; Rothmaier, M. and Flükiger, M. (2004)
Production of pharmaceutical compounds with Wave 20 SPS. In: Conference proceedings,
BioPerspectives, Wiesbaden (FRG), May 4-6 2004; pp. 338.
[19] Flores, H.E.; Hoy, M.W. and Pickard J.J. (1987) Secondary metabolites from root cultures. Trends
[20] Wilson, P.D.G. (1997) The pilot-scale cultivation of transformed roots. In: Doran, P.M. (Ed) Hairy
Roots: Culture and Applications. Harwood Academic Publishers, The Netherlands; ISBN 90-5702-117X; pp. 179-190.
[21] Kim, Y.H. and Yoo, Y.J. (1993) Development of a bioreactor for high density culture of hairy roots.
Biotechnol. Lett. 7: 859-862.
[22] Nuutila, A.M.; Toivonen, L. and Kauppinen, V. (1994) Bioreactor studies of Catharanthus roseus:
comparison of three bioreactor types. Biotechnol. Lett. 8: 61-66.
[23] Hitaka, Y.; Kino-oka, M.; Taya, M. and Tone, S. (1999) Effect of liquid flow on pigment formation of
red beet hairy roots. J. Chem. Eng. Jpn. 32: 370-373.
[24] Honda, H.; Liu, C. and Kobayashi, T. (2001) Large-scale plant micropropagation. Adv. Biochem. Eng.
[25] Hu, W.W. and Zong J.J. (2001) Effect of a bottom clearance on performance of airlift bioreactor in highdensity culture of Panax notoginseng cells. J. Biosci. Bioeng. 92: 389-392.
[26] Kieran, P.M.; Malone, D.M.; MacLoughlin, P.F. (2000) Effects of hydrodynamic and interfacial forces
on plant cell suspension systems. In: Scheper, T.; Schügerl, K. and Kretzmer, G. (Eds.) Advances in
Biochemical Engineering Biotechnology; Vol 67, Influence of Stress on Cell Growth and Product
Formation. Springer Verlag, Berlin Heidelberg; ISBN 3-540-66687-7; pp. 141-177.
[27] Tescione, L.D.; Ramakrishnan, D. and Curtis, W.R. (1997) The role of liquid mixing and gas-phase
dispersion in a submerged, sparged root reactor. Enz. Microb. Technol. 20: 207-213.
[28] Williams, G.R.C. and Doran, P.M. (2000) Hairy root culture in a liquid-dispersed bioreactor:
Characterization of spatial heterogeneity. Biotechnol. Prog. 16: 391-401.
[29] Zhong, J.J.; Pan, Z.W.; Wu, J.; Chen, F.; Takagi, M. and Toshiomi, Y. (2002) Effect of mixing time on
taxoid production using suspension cultures of Taxus chinensis in a centrifugal impeller bioreactor. J.
Biosci. Bioeng. 94: 244-250.
[30] Henzler, H.J. (2000) Particle stress in bioreactors. In: Scheper, T.; Schügerl, K. and Kretzmer, G. (Eds.)
Advances in Biochemical Engineering Biotechnology; Vol 67, Influence of Stress on Cell Growth and
Product Formation; Springer Verlag, Berlin Heidelberg; ISBN 3-540-66687-7; pp. 38-82.
[31] Lübbert, A. (2000) Bubble column bioreactors. In: Schügerl, K. and Bellgardt, K.H. (Eds.) Bioreaction
Engineering: Modelling and Control. Springer-Verlag, Berlin Heidelberg; ISBN 3-540-66906-X; pp.
247-273.
[32] Reuss, M.; Schmalzriedt, S. and Jenne, M. (2000) Application of computational fluid dynamics (CFD) to
modelling stirred tank bioreactors. In: Schügerl, K. and Bellgardt, K.H. (Eds.) Bioreaction Engineering:
Modelling and Control. Springer-Verlag, Berlin Heidelberg, ISBN 3-540-66906-X; pp. 208-246.
[33] Knevelman, C.; Hearle, D.C.; Osman, J.J.; Khan, M.; Dean, M.; Smith, M.; Aiyedebinu, A. and Cheung,
K (2002) Characterisation and operation of a disposable bioreactor as a replacement for conventional
steam in place inoculum bioreactors for mammalian cell culture processes. ACS Poster, Lonza Biologics,
SL1 4DY (UK).
225
www.taq.ir
R. Eibl and D. Eibl
[34] Rhiel, M. and Eibl, R. (2004) Der Wave als System zur Prozessentwicklung für Proteinexpressionen In:
Conference proceedings, Biotech 2004, Wädenswil (CH), May 11-12 2004; pp. 20.
[35] Singh, V. (2004) Overview of the Wave Bioreactor system. http://www.wavebiotech.com (accessed 17
July 2004).
[36] Singh, V. (1999) Disposable bioreactor for cell culture using wave-induced agitations. Cytotechnol. 30:
149-158.
[37] Lisica, S. (2004) Energieeintrag in Wave-Bioreaktoren. Modelling approaches. University of Applied
Sciences Wädenswil, Wädenswil.
[38] Eibl, R.; Eibl, D.; Pechmann, G.; Ducommun, C.; Lisica, L.; Lisica, S.; Blum, P.; Schär, M.; Wolfram,
L.; Rhiel, M.; Emmerling, M.; Röll, M.; Lettenbauer, C.; Rothmaier, M. and Flükiger, M. (2003)
Produktion pharmazeutischer Wirkstoffe in disposable Systemen bis zum 100 L Massstab, Teil 1, KTIProjekt 5844.2 FHS. Final Report; University of Applied Sciences Wädenswil, Wädenswil.
[39] Pechmann, G G.(2002) Disposable Wirkstoffproduktion im Wave-Reaktor mitanimalen
Suspensionszellen. Diploma Thesis, University of Applied Sciences Anhalt, Köthen.
[40] Kallupurackal, J. (2004) Beitrag zur Beschreibung des Energieeintrages im Wave-System. Semester
Thesis; University of Applied Sciences, Wädenswil, Wädenswil.
[41] Sowana, D.D.; Williams, D.R.G.; Dunlop, E.H.; Dally, B.B.; O`Neill, B.K. and Fletcher, D.F. (2001)
Turbulent shear stress effects on plant cell suspension cultures. Trans. I.Chem.E. 79: 867-875.
[42] Tanaka, H. (1981) Technological problems in cultivation of plant cells at high density. Biotechnol.
Bioeng. 23: 1203-1218.
[43] Dunlop, E.H.; Namdev, P.K. and Rosenberg, M.Z. (1994) Effect of fluid shear forces on plant cell
suspensions. Chem. Eng. Sci. 49: 2263-2276.
[44] Chen, S.Y. and Huang, S.Y. (2000) Shear stress effects on cell growth and L-DOPA production by
suspension culture of Stizolobium hassjoo cells in an agitated bioreactor. Bioprocess Eng. 22: 5-12.
[45] Präve, P.; Faust, U.; Sittig, W.; Sukatsch, D.A. (1994) Handbuch der Biotechnologie. 4. Auflage, R.
Oldenbourg Verlag, München Wien; ISBN 3-486-26223-8; pp. 189.
[46] Griffiths, J.B. (1999) Mammalian cell culture reactors. In: Flickinger, S.W. and Drew, S.W. (Eds)
Encyclopaedia of Bioprocess Technology, Vol 3, Fermentation Biocatalysis and Bioseparation. John
Wiley & Sons, Inc., New York; ISBN 0-471-13822-3; pp. 1594-1607.
[47] Eibl, R.; Hans, D.; Lettenbauer, C. and Eibl, D. (1999) Einsatz eines Taumelreaktorsystems mit interner
Beleuchtung. BioWorld 2: 10-12.
[48] Kato, A.; Shimizu, Y. and Nagai, S. (1975) Effect of initial kLa on the growth of tobacco cells in batch
culture. J. Ferment. Technol. 53: 744-751.
[49] Meijer, J.J.; ten Hoopen, H.J.G.; Luyben, K.C.A.M. and Libbenga, K.R. (1993) Effects of hydrodynamic
stress on cultured plant cells: A literature survey. Enz. Microbiol. Technol. 15: 234-238.
[50] Wheathers, P.J.; Wyslouzil, B.E. and Whipple M. (1997) Laboratory-scale studies of nutrient mist
reactors for culturing hairy roots. In: Doran, P.M. (Ed.) Hairy Roots: Culture and Applications. Harwood
Academic Publishers, The Netherlands; ISBN 90-5702-117-X; pp. 191-199.
[51] Yoshikawa, T. (1997) Production of ginsenosides in ginseng hairy root cultures. In: Doran, P.M. (Ed.)
Hairy Roots: Culture and Applications. Harwood Academic Publishers, The Netherlands; ISBN 90-5702117-X; pp. 73-79.
[52] Kato, Y.; Honda, H.; Hiraoka, S.; Tada, Y.; Kobayashi, T.; Sato, K.; Saito, T.; Nomura,T. and Ohishi, T.
(1997) Performance of a shaking vessel-type bioreactor with a current pole. J. Ferment. Bioeng. 84: 6569.
[53] Honda, H.; Hiraoka, K.; Nagamori, E.; Omote, M.; Kato, Y.; Hiraoka, S. and Kobayashi, T. (2002)
Enhanced anthocyanin production from grape callus in an air-lift type bioreactor using a viscous
additive-supplemented medium. J. Biosci. Bioeng. 94: 135-139.
[54] Kim, D.J. and Chang, H.N. (1990) Enhanced shikonin production from Lithospermum erythrorhizon by
in situ extraction and calcium alginate immobilization. Biotechnol. Bioeng. 36: 460-466.
[55] Brodelius, P. (1985) The potential role of immobilisation in plant cell biotechnology. Trends Biotechnol.
3: 280-285.
[56] Dörnenburg, H. and Knorr, D. (1995) Strategies for the improvement of secondary metabolite production
in plant cell cultures. Enz. Microbiol. Technol. 17: 674-684.
[57] Kieran, P.M.; MacLoughlin, P.F. and Malone D.M. (1997) Plant cell suspension cultures: some
[58] Warlies, S.; Reinhardt, K. and Eibl, R. (1999) Optimierung der Synthese von Alliin/Allicin mit
pflanzlichen Zellkulturen in vitro und im Bioreaktor. Report; University of Applied Sciences Wädenswil,
Wädenswil.
226
www.taq.ir
[59] Hirasuna, T.J.; Pestchanker, L.J.; Srinivasan V. and Shuler, M.L. (1996) Taxol production in suspension
cultures of Taxus baccata. Plant Cell Tissue Org. Cult. 44: 95-102.
[60] Ketchum, R.E.B. and Gibson, D.M. (1996) Paclitaxel production in cell suspension cultures of Taxus.
[61] Navia-Osorio A.; Garden, H.; Cusidó, R.M.; Alfermann, A.W. and Piñol, M.T. (2002) Taxol® and
baccatin III production in suspension cultures of Taxus baccata and Taxus wallichiana in an airlift
bioreactor. J. Plant Physiol. 159: 97-102.
[62] Cusidó, R.M.; Palazón, J.; Bonfill, M.; Navia-Osorio, A.; Morales, C. and Piñol, M.T. (2002) Improved
paclitaxel and baccatin III production in suspension cultures of Taxus media. Biotechnol. Prog. 18: 418423.
[63] Oksman-Caldentey, K.M.; Vuorela, H.; Strauss, A. and Hiltunen, R. (1987) Variation in the tropane
alkaloid content of Hyoscyamus muticus plants and culture clones. Planta Med. 53: 349-354.
[64] Ouhikainen, K.; Lindgren, L.; Jokelainen, T.; Hiltunen, R.; Teeri, T.M. and Oksman-Caldentey, K.M.
(1999) Enhancement of scopolamine production in Hyoscyamus muticus L. hairy root cultures by genetic
engineering. Planta Med. 208: 545-551.
227
www.taq.ir
PART 3
MECHANIZED MICROPROPAGATION
www.taq.ir
INTEGRATING AUTOMATION TECHNOLOGIES WITH COMMERCIAL
MICROPROPAGATION
An economic perspective
CAROLYN J. SLUIS
Tissue-Grown Corporation, 6500 Donlon Road (PO 702), Somis,
California 93066, USA - Fax: 805-386-8227Email:[email protected]
1. Introduction
Replacement of the people who do micropropagation work in laminar flow hoods, with
equipment of any kind, is neither technologically simple nor readily economically
achievable. The fundamental fact remains that the human eye-hand-brain combination is
both highly sophisticated, technologically, and incredibly inexpensive, certainly when
considered on a global scale. Consequently, commercial micropropagation companies
in both Europe and North America have followed the path of lower costs to those
countries for which the infrastructure, such as reliable power supplies, and logistics,
such as political stability and transportation issues, are favourable.
Cost accounting needs to take into consideration many factors which are not always
obvious at the onset of a project; in the case of micropropagation these include risk
assessments, refinement of protocols, and employee training. Aseptic culture systems
are vulnerable to bacterial, fungal and even insect contaminants which can destroy the
plantlets, as well as to genetic and epigenetic shifts, which can seriously impair their
quality. The transfer of the operation from human to mechanical means can
differentially affect each of these factors. The costs of maintaining a high level of
genetic purity and the risk of contamination must be factored into the long-term costs of
mechanized systems. The history of micropropagation has created a legacy of sudden,
disastrous plantlet losses, the magnitude of which have cooled the ardor of all but the
hardiest researchers. Likewise, the financing and funding of various companies and
projects has been erratic, often resulting in a lack of continuity and instability; as
evidenced by ventures such as Plant Genetics, of the United States, based on scale-up of
somatic embryogenesis [1], ForBio, of Australia, focussed on elite tree
micropropagation using robotics [2] and Osmotek, of Israel, a supplier of plastics for
biofermentation and liquid culture systems.
231
© 2008 Springer.
www.taq.ir
www.taq.ir
C.J. Sluis
2. Biological parameters
2.1. THE PLANT’S GROWTH FORM AFFECTS MECHANIZED HANDLING
The list of plants which can be grown in vitro is broad and covers many genera [3,4];
nonetheless, the vast majority of plant species are not able to be economically
micropropagated, due either to technical difficulties in tissue culture or to their expense
relative to standard propagation by seed, cuttings, tubers or bulbs. The growth form of a
plant can be significantly modified in vitro by the use of plant growth regulators and
environmental controls, so that a plant which normally grows as a linear vine, such as a
potato, can become either a linear, straight-stemmed plantlet, such as is commonly seen
in test tubes, or a dense compact cluster of buds, such as is possible in liquid culture
using ancymidol, or even a linear microtuber, as can occur under certain environmental
conditions. Other genera naturally grow as rosettes, and inducement of axillary bud
growth results in dense masses of tiny shoots (see the Limonium plantlets in Figures 1
and 2). For example, in the case of potato and carnation, the preferred growth form for
robotic access and separation has been linear; the plantlets can be grown upright, then
laid flat for cutting (see Figure 3) or they can be grown in shallow plastic boxes with
domed tops and repeatedly “hedged,” as was described by Aitken-Christie and Jones for
pine [5]. Both of these methods are effective in increasing access for mechanical
handling of the plantlets. Potatoes can be grown in liquid culture as nodes [6-8] as bud
clusters [9-11] or made into microtubers [6,8,12,13] or even somatic embryos [14].
However, none of these methods have been scaled up to millions of plantlets due to two
barriers:
x Economics: the potato industry is based on tuber seed pieces costing less than
a penny a piece
x Size issues: in North America, field conditions dictate that the tuber seed piece
will not be replaced by anything smaller than a greenhouse minituber for many
years to come.
Figure 1. Axillary branching in statice using liquid medium additions.
232
www.taq.ir
Integrating automation technologies with commercial micropropagation
Historically, commercial micropropagation was based on enhanced axillary bud break;
overcoming the natural apical dominance with cytokinins and other factors to encourage
lateral buds to grow out into shoots; this increases the number of shoots per culture per
month, the multiplication rate. However, if strategies with lower multiplication rates, for
example, straight stemmed, unbranched shoots, give significant advantages to
mechanization, then branching options may been to be reexamined within the new
framework. Multiplication rates of greater than 10-fold per month can be achieved in
tissue culture (see Figures 1 and 2); however, these may not contribute significantly to
reduce the end-product cost if the labour for singulation/rooting is increased (see Table
1) and/or the quality of the plantlets decreased.
Figure 2. Subculture of explants (shown in Figure 1) for rooting.
Figure 3. Cassette style (square Petri dish) of potato cultures at 2 week for robotic access.
233
www.taq.ir
C.J. Sluis
Somatic embryogenesis continues to be a highly attractive biological strategy for largescale production research, despite the difficulties, although full automation still appears
to be years away. The largest ongoing operations based on this technology appear to be
in the forestry sectors, where manual handling of the output embryos is still the norm,
dramatically raising final costs [15]. Even if the embryos cost next to nothing apiece
and can be made in the hundreds of thousands, a single manual handling step, such as
singulation or planting, can make the system economically prohibitive [16]. The genetic
component within species regarding the ability to form somatic embryos can be
significant [14] as is the potential for loss of genetic fidelity in the pre-embryoid tissues.
Genetic testing technologies are assisting vegetable breeding companies in confirming
the true-to-type characteristics critical for seed parents (Rijk Zwaan, personal
communication) and the existing automation of PCR testing of cotyledon discs could
enable future monitoring of somatic embryo-derived plugs; however, the automated
somatic seed concept still appears many years off [17,18].
On an international scale, another biological method for micropropagation, known
as bud clusters, has gained widespread acceptance both for its potential application to
many species, as well as its obvious physical compatibility with mechanical handling
[19]. One method developed by the late Levin [20] and Ziv [11,21,22], combines the
bud cluster growth form, either in liquid culture or on agar, with a simple fixed-blade
mechanical cutting device, such as a grid of blades, allowing the clusters to be
mechanically subdivided into up to 100 pieces with one operation; this has been shown
to work in potato, lilies and several other crops.
The most common media typically used for induction of bud cluster growth patterns
involve the use of liquid culture, a gibberellin inhibitor, such as ancymidol, and an
axillary bud growth promoting agent, such as the cytokinin benzylaminopurine. The
bud cluster induction treatment needs to be repeated serially for several subcultures to
establish the formation of true clusters. It is difficult to scale up to commercial levels in
liquid culture systems, due to hyperhydricity [21] and bacterial contamination problems.
While this avenue of production research has great potential for long-term production in
high volumes of quality plantlets, the difficulties remain problematic and the
limitations, especially for commercial laboratories, remain significant. Several major
genera of plants already in mass propagation via tissue culture are quite amenable to the
liquid bud cluster construct, as they readily form a densely compact mass of basal
proliferation and are tolerant of high humidities, liquid environments and mechanical
damage. These will probably be propagated in increasing numbers over time using
biofermentation approaches.
Researchers in several crops and from several countries are scaling up the bud
cluster system [11,19,21-24]. Basically, cluster culture involves the reduction of the
tissue culture plantlet to a compact mass of leafless, highly branched, short masses of
buds; there is little or no callus proliferation or adventitious bud formation. These
organized bud clusters are then maintained in a multiplication mode as long as
necessary for production of sufficient numbers to meet the goals of the project. When it
is time for finishing the plantlets, the pressure of the cytokinin/growth retardant
combination is removed and the shoots grow out into their normal morphology.
234
www.taq.ir
Currently the most advanced commercial biofermentation systems in application are
based on the incubation of cultures in a redesigned biofermentation vessel consisting of
a five or ten liter autoclavable plastic bag, similar to a medical medium or serum bag,
complete with input and output ports. Implementation of this technology is being
intensely pursued by at least two major high volume laboratories in North America.
Rather than being an automated system, biofermentation of bud clusters is actually still
an operator-assist method, the subcultures are still carried out by hood operators; the
vessels combined with bud clusters greatly increases the productivity of the operator
and hence significantly reduces the cost per plantlet, while still benefiting from human
decision-making. Attempts to automate these systems further have not yet been realized,
but are nearing. The bag fermentors, equipment and supplies facilitating the production
of plantlets, bud clusters, somatic embryos and other propagules in liquid fermentation
was commercially available prior to 2004, but the withdrawal of the manufacturer
currently makes the development of liquid bud cluster systems less accessible to smaller
operations.
Liquid culture systems are, in general, more difficult to stabilize, maintain and
commercialize than simple agar-based standards. Humidity must be carefully managed
for maintenance of consistent medium volumes and component concentrations. In
smaller vessels, the variability between vessels can be dramatic. When propagation is
transferred from agar-based to liquid many factors in the medium itself will need to be
adjusted. In some cases this can amount to starting from scratch, never an attractive
option for the tissue culture propagation laboratory.
Many plants do not take easily to being submerged in liquid. To overcome problems
such as hyperhydration, deformed growth, insufficient cuticles and other side effects of
oxygen depletion and underwater growth, enhanced oxygenation of the solution, and
timed, temporary immersion rather than full-time exposure to the liquid environment
can improve the quality of plantlets substantially [25,26]. However, intermittent
flooding, while clearly of benefit to many species, is cumbersome and even more prone
to difficulties with contamination, so challenges remain [21].
2.2. MICROBIAL CONTAMINANTS HINDER SCALE-UP
Microbes present much more of a threat to the mass propagation of plants in vitro than
they do in greenhouses. Normally harmless, airborne organisms, such as molds, yeasts
and otherwise unheard of bacteria [27-31], become lethal to plantlets in the
micropropagation environment, simply by overwhelming the cultures. Internal bacteria,
some of which can be quite significant, can be carried at extremely low populations for
years without detection [32].
Plant tissue culture originated in tightly capped, glass culture vessels using very
small tissues, such as meristems, which had no capacity to produce sufficient
photosynthate for growth and development. Consequently, sugars were required in the
medium, and sugars are used in nearly all of today’s commercial laboratories, including
our own. Plants do not normally require extraneous sugar for growth and development;
the artificial conditions of restricted gas exchange, low light levels and high humidity,
incur the need for sugar in tissue culture media. While true meristems, embryos,
protoplasts and other tissues certainly require carbohydrate sustenance; micropropagated
235
www.taq.ir
C.J. Sluis
plantlets are fully capable of supporting themselves. The micropropagation industry has
paid heavily for its reliance on sugar, both from the severe restrictions on automation
and mechanization research resulting from the extreme requirements for sterility in any
process involving sugar-based production, and from the plantlet losses during
transitioning due to weaknesses in the epidermal tissues and root systems [33-38].
Photoautotrophy has clearly been demonstrated to produce healthy and vigorous plants,
but it has not been fully incorporated into production laboratories.
Photoautotrophy, which clearly reduces the bloom of microorganisms and which
equally clearly promotes healthy plantlet growth, has not been an easy goal to attain at
the commercial level, in part due to the reluctance to aerate the culture vessels, thereby
risking contamination, which can be a very real problem, and in part due to reluctance
to spend significant funds on facilities and culture vessel modifications. The
requirements for environmental controls and modified vessels are somewhat stringent in
order to achieve true parity on a production scale. Cutting corners, while still permitting
improvements in plant performance, do not help with bacterial control in automation
research, as even a little sugar in the medium will support very vigorous microbial
populations. Although green plantlets conduct photosynthesis while in tissue culture,
the rates are often low and reliance on sugar is high, even in the greenest plantlets. Still,
it is logical that photoautotrophy or at least enhanced photomixotrophy [25,38] will
become standard for standard types of commercial propagation in vitro.
Culture indexing, whereby plantlets or tissues are assayed for the presence of
internal, or non-obvious, bacteria is commonly practiced using several standard media
which encourage bacterial growth, such as nutrient broth and potato dextrose agar.
While culture indexing is important in agar-based systems, it is critical for liquid-based
systems, where contamination can overtake the cultures within a matter of days, or even
hours.
Sterility is critical to maximum batch size, as a greater percentage of the plantlets
produced are at risk when more explants are in a single vessel. Obviously, if plantlets
are subcultured in test tubes, and 1% of the explants are contaminated, then 1% of the
plantlets are lost; however, if 50 plantlets are subcultured into each culture vessel, a 1%
contamination rate quickly adds up to many more plantlets being lost.
Antibiotics and bactericides, such as hydrogen peroxide and sodium hypochloride,
have been added to culture media to kill bacteria, or at least inhibit their growth [30].
Other strategies, such as refrigeration, filtration or ozonation of the recirculation
medium, have been implemented to a lesser degree [39,40].
3. Physical parameters
Several physical parameters can be re-examined for potential modifications or options
which may favour new automation or mechanization technologies. Physical constraints
which have been accepted as fixed for standard parameters may need to be modified in
order to make new systems feasible. For example, the benefits of automation on final
cost-per-unit may ultimately outweigh the subsidiary input costs of using more
expensive culture vessels. The benefits of photoautotrophy may outweigh the outlay of
expenses for culture room modifications.
236
www.taq.ir
3.1. CULTURE VESSELS
The physical parameters of the micropropagation system begin with the choice of
culture vessel. The culture vessel either permits ready access or hinders it; it allows
varying degrees of gas exchange and clarity, and it has an impact on plantlet growth and
quality. Many factors come into play when choosing a vessel for commercial
propagation. Inexpensive culture vessels which impede operators are, in fact, far more
costly than slightly more expensive culture vessels which streamline labour. From a
materials-handling perspective, glass is heavy, awkward and requires washing, an added
expense. From an access perspective, test tubes are seriously limiting, and operators can
rarely handle more than 800 per day; but test tubes retain their usefulness in many
applications, including culture initiations and germplasm maintenance. Culture vessels
may be designed specifically with an automation device in mind, as is the case with
most robotic applications [41,42].
The choice of culture vessel is also important to controlling contamination losses:
the larger the vessel, the greater the number of plantlets which are lost with each
introduced contaminant. Consequently, the use of larger vessels typically requires ultraclean laboratories, incurring additional facilities costs [43]. In addition to the higher
multiplication rates attainable in 10 L liquid culture bags, these vessels have good
accessibility throughout the subculture cycle, and daily operator productivity, can be
increased substantially as a result.
3.2. PHYSICAL ORIENTATION OF EXPLANTS FOR SUBCULTURE OR
SINGULATION
Over the past 20 years, many different concepts for the mechanization or automation of
micropropagation have been envisioned; originally, mechanical approaches were based
on either robotics with computer imaging, for cutting of straight stemmed cultures
(potatoes, trees, sugarcane, carnations) [41,42], or adventitious regeneration approaches,
which are combined a ‘blender’ approach to cutting of tissues, with species such as
ferns. Subsequently, researchers studied the semi-automated production of artificial
seeds using somatic embryos [2,7,18].
Each of these systems had its drawbacks and limitations. For mass regeneration
systems, the phenotypic and genotypic changes of somatic embryos were problematic in
crops which required a high degree of uniformity [17]. For robotic cutting systems,
there were few suitable crops needed in the volumes required to amortize the high costs
of the initial production line and its maintenance, and there were ongoing issues of low
speed relative to the human operator. In the case of somatic seed, commercial efforts
still had a heavy reliance on operators at the final stages of singulation and sorting.
Bud clusters are physically compatible with random, or spatial, mechanical cutting
equipment in the multiplication stages, as there are so many buds in various stages of
development that damage to a certain percentage of them is bearable. Once true bud
clusters have been created, subdividing the clusters by means of mechanical, fixed blade
cutting devices becomes feasible [9,22,24]. For potatoes, even operator-assist devices,
such as grid blades (similar to French fry cutters) can greatly increase efficiency, as
essentially 25-36 sub-divisions can take place with one cut. Resterilization of the grid
237
www.taq.ir
C.J. Sluis
blades over the course of the day is not any more cumbersome than resterilization of
forceps and scalpels, but the cost of multiple tools and handling the cutting devices is
slightly more expensive and awkward.
3.3. GAS PHASE OF THE CULTURE VESSEL IMPACTS AUTOMATION
Plantlets grown under conditions of reduced humidity, reduced ethylene, adequate
carbon dioxide and adequate oxygen perform better during the transitioning period,
which is instrumental to elimination of the tissue culture rooting stage. The choice of
vessel influences the amount of gas exchange possible between the sterile interior and
ambient, or external air. Currently, biofermentation using temporary immersion or
nutrient film delivery techniques, rather than full submersion, can provide environments
that are highly favouable to the plantlet in terms of both photosynthetic activity and
epidermal function.
Innovations in photoautotrophy are accompanied with greater understandings of the
effects of light spectra and intensity on the quality of plantlets [44-46]. Research into
“chopper light” may allow significant savings in cooling costs, as well as decrease
electrical costs for lighting.
Greenhouse operations have been adding carbon dioxide to the plant environment
for years. Increased carbon dioxide in the growth room (at 2-4 x ambient levels) can
enhance the performance of plantlets even on sugar-based media, especially when
culture vessels are well vented. Advances in porous filters and tapes (i.e. 3M
Micropore™ tape) have enabled the venting of many previously sealed containers.
4. Economic parameters
For any new technology, such as automation of micropropagation, the primary indicator
of its commercial potential is its projected impact on the cost of the plantlet. While true
cost accounting is a complex and multifaceted task that is required for ongoing
operations and fine decision making [47], it can be simplified for the purposes of
preliminary evaluations. For this purpose Table 1 was designed to permit comparison of
various factors, such as labour daily costs and multiplication factors; it is a model only,
each major crop group within each commercial laboratory requires its own analysis for
accurate cost accounting.
4.1. BASELINE COST MODELS
The total payroll of micropropagation laboratories is typically over 65% of the monthly
budget; however, this does not give an accurate picture of the pyramid of costs linked to
each hood operator hour. Costs need to take into account all aspects of the operation, so
one simplistic approach, used by several laboratories including ours, is to take the total
monthly outlays and divide them by the parameter being evaluated, for example hood
operator hours per month (excluding medium preparation, dishwashing and other nonhood activities), for an average cost per hour of the hood work.
238
www.taq.ir
Table 1a. Model of cost per plantlet, as influenced by various factors.
Multiplication
Rate (xx)
TC Systems:
Daily hood operator rates
600/day
900
Standardb
1200
1500
1800
2100
2400
Cost fixed at $35/hr fully loadeda
2x
$0.933
$0.622
$0.467
$0.373
$0.311
$0.267
$0.233
3x
$0.700
$0.467
$0.350
$0.280
$0.233
$0.200
$0.175
4x
$0.622
$0.415
$0.311
$0.249
$0.207
$0.178
$0.156
5x
$0.583
$0.389
$0.292
$0.233
$0.194
$0.167
$0.146
6x
$0.560
$0.373
$0.280
$0.224
$0.187
$0.160
$0.140
7x
$0.544
$0.363
$0.272
$0.218
$0.181
$0.156
$0.136
8x
$0.533
$0.356
$0.267
$0.213
$0.178
$0.152
$0.133
9x
$0.525
$0.350
$0.263
$0.210
$0.175
$0.150
$0.131
10x
$0.519
$0.346
$0.259
$0.207
$0.173
$0.148
$0.130
20x
$0.491
$0.327
$0.246
$0.196
$0.164
$0.140
$0.123
30x
$0.483
$0.322
$0.241
$0.193
$0.161
$0.138
$0.121
40x
$0.479
$0.319
$0.239
$0.191
$0.160
$0.137
$0.120
50x
$0.476
$0.317
$0.238
$0.190
$0.159
$0.136
$0.119
60x
$0.475
$0.316
$0.237
$0.190
$0.158
$0.136
$0.119
70x
$0.473
$0.316
$0.237
$0.189
$0.158
$0.135
$0.118
Advancedc
a
Fully loaded cost per hour includes both direct and indirect costs: facilities, utilities, materials, freight
Standard tissue culture (TC) includes: axillary branching, nodal culture
c
Advanced tissue culture (TC) includes: somatic embryos, adventitious bud cultures, hedge, biofermentation
b
239
www.taq.ir
C.J. Sluis
Table 1b. Variation in plantlet cost with global labour costs.
Loaded
cost per
hour
(US$)
TC
Systems:
Daily hood operator rates
600/day
900
1200
1500
1800
2100
2400
$35/h